Analog
Logic
RCA Power Transistor Applications Manual (1983) has descriptions hometaxial, etc.
Cross-evaporation: US 3709695, Bowman, R., "Fabrication of Semiconductor Devices", issued 1973-01-09, assigned to Motorola
The Costas loop article needs work. See Talk:Costas loop.
The angle addition formulas:[1]
We can use those formulas to find the result of a phase shift.
Angle addition formulas taken together produce the heterodyne formulas:
These clarify the first image in the article that has a simple story of in phase and quadrature. The diagram uses a sin() phase carrier and a sin() phase VCO.
The image does not give the heterodyne formulas, but the image shows a zero phase difference between the carrier and VCO because both are described as . To model the phase difference, let a be the carrier phase and b the VCO phase. The upper and lower branches then produce:
The summed phases get filtered out. Demodulation is on the upper branch: cos(a-b) is approximately cos(0) = 1. Phase control on the lower branch. The lock point is stable at near zero phase difference: if b gets a little ahead of a, then sin(a-b) is negative which will reduce the VCO phase.
This diagram started the article, has the cleanest math, and the rest of the article should follow suit. Top branch should be demodulation and bottom branch phase control. The phases going to the mixers should be sin() and cos().
In contrast, the two diagrams in the final section of the article are confused, inconsistent, and changing.
The recent formulas given in the diagrams for the lower branches are:
We should get the lower formula by substituting for , but that gives
Notice the -θs; that makes the lock point stable for small values. This last formula is identical to the cos case.
In other words, the formulas says the first diagram locks with a phase difference near 0, and the second diagram locks with a phase near 180°.
If we look at earlier versions of the two diagrams, the situation is still confused. The "before sync" SVG file from 2012 recognizes that the -90° phase shifter creates a minus cosine, so it negates m(t). That makes the loop unstable for small phase differences between the carrier and the VCO. The "after sync" SVG from 2012 is even more confused. It assumes the phase difference between the VCO's sin() and the carrier's sin() is zero. That is evidenced by the sin(0) expression in the lower branch. It expects the top branch to provide the demodulated m(t), so it is trying to mimic the diagram in the first section. It does not show the phase error, so we cannot comment on stability.
The same issues are present in another diagram:
The expressions for the upper and lower branches with a −90° phase shift to the lower branch should be (with a the carrier phase and b the VCO phase):
And -sin(-x) = sin(x) by antisymmetry, so the "after lock" lower branch formula is correct, but the lock is unstable as shown. IF the VCO is a little fast, then the phase difference with be positive, and the feedback will speed up the VCO even more. The diagram is wrong (assuming positive filter gain).
Amelia Earhart#Radio equipment
Is a confusing mess.
Cyril D. Remmlein is in Bendix Radio Research Division. In February 1937, Remmlein brought a prototype Bendix Receiver to Newark, NJ airport. The first receiver was still in manufacturing. O. Vernon Moore, the design engineer, had not designed the coils yet. Earhart and Manning choose 200 to 10,000 kHz in 5 bands. They would get the first production receiver; it would be shipped to Burbank when ready.[1]
Joseph Gurr is UA radio tech at Burbank. Comes on the scene because WeCo radio did not work. Gurr quickly discovers antenna not plugged in. "From then on, Gurr was the person they called whenever they needed a radioman."[2] Become go-to radio guy for Earhart and Putnam. Gurr had nice comments: you must do the last hundred miles on radio. https://tighar.org/wiki/Modifications_by_Joe_Gurr
Gillespie says plane shipped with starboard ventral antenna and a trailing wire antenna from the tail. https://tighar.org/wiki/NR16020_antennas#Chronology Pitot tube to two ventral masts. https://tighar.org/Publications/TTracks/1995Vol_11/starborn.pdf says TA is just below tail light on delivery; also says not the TA that was removed; comment at end of article about pinning down nomenclature of the loop.
October 1936 has installation of Hooven.
Jan-Feb 1937 had installation of Vee. (I think this was a Bell Labs installation; they could reasonably install the WeCo receivers.)
However, probably installation of Bendix RA-1 receiver. Who is Remmlein? He flew out from Washington DC to do the modification.
Gas tanks installed (limiting access to tail). Also during that time, trailing wire antenna moved to ventral mast underneath cabin.
Somewhere, the trailing wire has a weight. That would incease vertical component of antenna and improve forward transmission. Long & Long (1999, p. 173): "Also installed was a 250-foot trailing wire antenna, which was weighted on the end and wound onto an electrically driven reel in the Electra's tail. Low frequencies, such as the 500=kilocycle channel, require long antennas and communicate in code only." The "in code only" is dubious; lower end of BC band is 550 kHz.
Just before flight, Bendix donates $2,500.
See 1941 Bendix Brochure for MN-31 Automatic Radio Compass.
Hooven removed in early March 1937. Gillespie says Hooven Dome replaced with Bendix MN-5 loop. Long and Long say the loop is a Navy RDF-1-B loop with a custom coupling box.[3]
TIGHAR probably has it right that dorsal was transmit-only and the transmitter lead went from transmitter to leg of vee.
Source: 1962 letter from W. C. Tinus, Vice President of Bell Telephone Laboratories:[4]
There was a reference on design of dorsal antennas. Expect Tinus did a good job; does TIGHAR have the dimensions? Then Gurr changed the length (and the tuning). Didn't work well on trip to Miami, so Miami changed it.
There's an issue of how Gurr switched the transmitter to the 500 kHz range, but that becomes irrelevant if Miami removed the network. Expect a direct connection with no additional loading/matching networks.
TIGHAR probably right here that just went to the receiver.
Could be used for sense antenna, but then lead would have to go up to the RDF unit.
TIGHAR wipe-off theory is probably right, too. Earhart's receiver still worked because she got the RDF receiver to work Howland. That means the fuse probably did not blow. Earhart could not get the 3 or 6 kHz channels to work, so the antenna was probably disconnected or wiped off. Alternatively, the controls may not be set properly, but Earhart had that working at Darwin.
There are more complicated theories about connection....
There are pictures of Earhart with the loop and coupling box. https://tighar.org/smf/index.php?topic=770.15 has pictures. Box has controls
Long & Long (1999, p. 63):
The loop coupler control box had been custom-made to electronically adapt the standard Navy RDF-1-B direction-finder loop to the new Bendix RA-1 receiver. Five frequency-band numbers were shown on the face of Earhart's loop coupler, but the Navy model had six bands. The actual frequency range of each band was listed on a chart concealed inside the inner bottom cover plate of the unit. The limited space on the face of Earhart's loop coupler required that the frequency bands be labeled 1, 2, 3, 4, and 5. Remmlein cautioned Manning that the radio direction-finder loop was designed to take bearings only on frequencies between 200 and 1430 kilocycles, whereas the receiver could receive signals on frequencies up to 10,000 kilocycles. If Earhart or Manning tried to take a bearing on a high-frequency signal using the Bendix receiver's upper bands, the resonance of the loop would be so far out of tune they probably could not get a minimum signal.
Next paragraph says 5 bands do not match.
Notes for page 63 cite to the Bendix RDF-2-A Navy instruction book, page 12, for "limited space". A 21 May 1976 interview of Al Hemphill, Bendix direction finder engineer, "If Earhart or Manning tried" the loop would be so far out of tune they could not get a minimum, and 20 January 1976 interview of Joe Gurr for "natural mistake" of matching band numbers.
RDF was useless flying into Africa. https://tighar.org/Projects/Earhart/Archives/Forum/Forum_Archives/199910.txt (apparently from Long and Long quoting letter from Noonan dated 6/8/1937.
Gillespie dubious of Long: "From what I've seen, Elgen presents no documentation to support most of his allegations about the Bendix DF and he completely neglects to mention the earlier Hooven/Bendix Radio Compass that was installed and later removed." (Forum Archives 199910.txt)
IIRC, the notes section of the book refers to the RDF manual (a later version that I did not find).
See description below.
R T D B
RDF-1B see http://aafradio.org/docs/Navy-radio-gear-1943.pdf
Bendix documents (Reference Drawing AR3000) definitively identify Earhart's RDF as an RDF-2-A with an L1678A coupler, but the units may be prototypes.
There's a picture of a later coupler that has operating instructions.[2] Much better picture at National Air and Space Museum: https://airandspace.si.edu/multimedia-gallery/4523hjpg Radio Compass, Navy, DU-1, Bendix. The loop is two-tone color.
R connects input omni antenna (terminal "A") to output (to receiver via J102). RDF unit may be off. B would add capacitors to resonate without adding omni. Omni input grounded; just loop operative. Loop tuning would be a variable capacitor to fine tune the loop. Variable capacitors do not have wide range (e.g., 40 to 360 pf is range of 3 for broadcast band). Loop range of DU-1 is 200 to 1600 kHz. That's 3 octaves. That's why the picture with instructions would have 3 bands. The bands are 200 to 400, 400 to 800, and 800 to 1600. Earhart's coupler may have used smaller range variable capacitor and therefore used 5 bands. D setting includes omni antenna to get a cardioid pattern. There's an on-off switch. DU1 uses two 12SK7 pentodes; power may be off in "R". http://www.r-type.org/pdfs/6sk7.pdf
There's some nice cleverness here. The loop has a fixed 200 pF capacitor (C-201). The tuning variable capacitor (C-102) is 19–395 pF. Autotransformers (T-101, T-102, and T-103) are used to boost the effective capacitance. For example, to get an octave range over the 200 pF fixed capacitor, one would need 0–600 pF. A transformer can lower the capacitors apparent impedance (increase its apparent capacitance). That gets around the limitation of the RDF-1A that switched fixed capacitors into the lower ranges, so the lower bands covered much less than an octave.
Gillespie quotes
They may be used as fixed-loop homing devices or as navigational direction finding instruments within frequency range of 200–1500 kcs.
((cite journal))
: |section=
ignored (help); not on Internet Archive.
See https://tighar.org/wiki/Radio_equipment_on_NR16020
Also https://tighar.org/smf/index.php?topic=452.120 where he offers a PDF.
Comments about second ventral antenna used as Hooven sense antenna.
https://tighar.org/Projects/Earhart/Archives/Research/Bulletins/53_MiamiPhoto/53_MiamiPhoto.htm Shows cockpit control box with an argument about a HF direction finder. But that photo fits with Long and Long description of RA-1 with remote control at copilot's station (sic pilot). Bendix RA-1B was used during the war. Manual: MR-1B remote control. Remote control uses manual speedometer cables for tuning and bandswitching.
Bendix Brochure has RA-1 pictures and MR-1 remote control. http://aafradio.org/docs/Bendix_1941_brochure.pdf
There is no magic RDF in the RA-1 receiver. RA-1B manual shows it as a conventional communications receiver. Its ability for RDF would be detecting minimum and maximum signals. The RA-1B has an antenna selector.
http://antiqueradios.com/forums/viewtopic.php?f=2&t=274814
Luke Field crash. Suggestion of port side underbelly damage. Did the ground loop damage the port side? Trailing wire and starboard ventral antenna were not reinstalled. There's some background, too. Manning left the flight after the crash.
Manning was the only one who could receive Morse code (15 wpm). Manning was the only one who had an FCC license to transmit Morse (obtained in March 1937). Manning was the only one who could transmit on 500 kHz (Earhart and Noonan were restricted radiotelephone licenses).
Consequently, a 500 kHz transmitting antenna would be surplusage. It may be that the RA-1 was viewed as a Morse accessory, too. That would have it pulled after first attempt but before second. Don't know if sources support that. Gurr did train Earhart on RA-1 in March according to Long and Long 1999.
Further conflict. March was before first attempt. Gurr modified dorsal antenna for 500 kHz. No need to do that unless trailing wire removed.
TIGHAR has this theory that lengthening the Vee created a horrible mismatch. At the same time, there's a comment that the feedpoint changed. Hello, offset dipole. Traps also popular.
Theory that during the Lae takeoff, the belly antenna was lost. Probably caused the loss of the Electra's ordinary receiving antenna.
Sources suggest RDF reception at 7500 kHz is very suspect. Could not tune the loop to that frequency. Loop might receive it, but trying to tune the loop would severely attenuate the signal. Indicates must be close. Relies on BFO for good reception; otherwise it is receiver quieting. If Earhart were using wrong antenna, then swinging the loop would have no effect.
Fundamental limit: λ = 300Mmps / 7.5 MHz = 40 meters. Give 4 meter for 0.1 λ. 10 inch loop is 1 meter, so 4 turns. Calculate inductance and resonate with 200 pF. Large loops on ships would be a problem. Where was the picture of those loops?
Lead in issue. Practice at the time (from RA-1B manual) was single wire lead in made short as possible. That would fit with Gillespie having transmitter in cabin near Vee antenna connection. Too much trouble to have ventral leadin go to roof to reach coupler and then come back to floor. Consequently, R is dysfunctional, T D B essentially identical. No unidirectional determination.
One modified WECo receiver. Is there an A/DF switch? How was TA connected. There was a LF antenna (bands 1 and 2) and HF antenna (bands 3 and 4). Band 1 188 to 420, Band 2 485 to 1200, Band 3 1500 to 4000, Band 4 4000 to 10000 (is this where Long believes RA-1 stops at 10 MHz?). https://tighar.org/Projects/Earhart/Archives/Research/ResearchPapers/ElectraRadios/Figure1.html So TA could couple into LF and ventral antenna to HF (or dorsal antenna with breakin). Must route HF through coupler because there is no DF setting.
One prototype Bendix RA-1 receiver. FA/TA/DF switch. Probably not remote controlled, so would have to go aft to flip the switch. Gurr enamored of receiver, but could he be enamored of WECo as well? Was Gurr asked in the 1970s? What if AE trained on RA-1 (use DF switch), but then RA-1 removed (no DF switch on WECo)?
Both receivers. Only search for null because receiving antenna would be connected to WECo.
Directionality of antennas. Fly around airport at Lae would be both close and broadside. Receiving antenna pointed at target would at null? Broadside would be collecting atmospherics.
Would turning plane onto sun line give directional increase?
Earhart did try receiving 500 kHz at Lae.
Only time Earhart determined a minimum was at Burbank under direction (but also flying to Hawaii). Gurr expected her to improve her skills during travels. But if in March, she was heading for trouble. Did RDF work going into Honolulu? (Advantage of high peaks; probably familiar to Gurr and Manning; Manning could operate equipment.)
Explanations for no minimum. If coupler has antenna connected and T not peaked, then omni could dominate and there would be no minimum.
Training for second flight. Yes, a lot more legs, but RDF failed going into Africa and may not have been repaired until Darwin. Likely that broken before Africa and not repaired until Darwin. If repaired before Darwin, then it broke again. Alternatively, AE failed to use equipment properly closing Africa; equipment broke after Africa but before Darwin; equipment likely to break again. AE watched while tech took bearing on ground; no indication of training AE. Darwin repair comments have been used to support 2nd receiver, but it may have just been the coupler. The failed generator is a twist. No fuse in couplers. Dynamotors may have had fuses. But coupler would have run off receiver dynamotor.
Long says remote for RA-1 at copilot station, but image shows in pilot station. (Miami photo).
Coupler designs. Reactance: 14154.281670205
The simple view. WECo transmitter works and is connected to dorsal Vee. Everybody hears 3105. 6210 reportedly has problems at Lae and nobody hears it after flight. Maybe Vee has problems at 6210. Only the WECo receiver. RDF connected to LF terminal and only used for bidirectional bearing. That means anything below 1400 is heard only through the loop, and that would be the right thing for RDF to radionavigation beacons and broadcast stations. It does require that the BFO be switched on for radionavigation beacons. Radio repair technicians at Lae probably took bearing on 500 kHz signal at Sal. while Earhart watched. Earhart failed to take bearing on 7500 at Lae because loop not connected to receiver in bands 3 and 4. She could turn loop all she wanted and not detect any change.
Earhart should have been able to take bearing on 500 kHz beacon transmitted by Itasca. With simple WECo connection, selecting tuning band would get correct antenna. But operator still needs to turn on RDF, select proper band and function (anything but R) on RDF coupler, turn on BFO, and turn off AVC. If fuse had blown again, then the loop would not work at all.
If ventral antenna wiped off at Lae, that would explain no HF reception. But it causes a huge problem explaining the 7500 kHz report. If there were two receivers, then the loop would always be connected to it, and they may have only tried to receive 7500 on that receiver (never looked for 3105 or 6210 or 500. Attempt would have been compounded by not using BFO. For 7500 kHz test at Lae, were they using voice? Failure to obtain a minimum implies receiving a signal; Earhart thought the signal was too strong. Unlikely for Lae to have 7500 kHz transmitter; did it have voice on nearby frequency?
If radio comm worked when RDF fuse was blown, then omni antenna not connected to RDF (one receiver model).
Earhart confused about equipment. Earhart thought whistling was necessary. If she did not do Morse code, then BFO would be foreign.
(Hey, switching to CW would disconnect Vee!)
Long and Long have nice section showing the Captain of the Itasca would have few worries because there were multiple ways for locating the Itasca.
Band | Radios | Radio Direction Finders | ||||||||
---|---|---|---|---|---|---|---|---|---|---|
20B (modified) |
RA-1(A) | RA-1(B) | RA-1B | RA-1J | RDF-1-A L1130A |
RDF-1-B ??? |
RDF-2-A (AE) L1678A |
RDF-2-A ??? |
DU-1 CRR-50061 | |
1 | 0.188 0.420 | .150 .??? | .150 .??? | .150 .315 | .150 .315 | 0.500 0.590 | ??? | ??? | .200 .??? | 0.200 0.400 |
2 | 0.485 1.200 | ??? | ??? | .315 .680 | .315 .680 | 0.590 0.700 | ??? | ??? | ??? | 0.400 0.800 |
3 | 1.500 4.000 | .??? 1.500 | .??? 1.500 | .680 1.500 | 0.680 1.500 | 0.700 1.050 | ??? | ??? | ??? | 0.800 1.600 |
4 | 4.000 10.000 | 1.800 ?.??? | 2.500 ?.??? | 1.8 3.7 | 2.500 5.000 | 1.050 1.570 | ??? | ??? | ??? | NA |
5 | NA | ??? | ??? | 3.7 7.5 | 5.000 10.000 | 1.570 3.000 | ??? | ??? | ?.??? 1.430 | NA |
6 | NA | ?.??? 15.000 | ?.??? 20.000 | 7.5 15.0 | 10.000 20.000 | 3.000 5.400 | ??? | NA | ??? | NA |
7 | NA | NA | NA | NA | NA | 5.400 8.000 | ??? | NA | NA | NA |
QUOTE FOR CUTTER AT HOWLAND ISLAND SIX HOURS AFTER DEPARTURE OF PLANE FROM HONOLULU CUTTER PLEASE CALL THE PLANE FROM 1 TO 6 MINUTES PAST THE HOUR AND THE FIRST 2 MINUTES ON 3105 KCS THE 2ND 2 MINUTES ON 6210 AND THE 3RD 2 MINUTES ON 500 KCS UNTIL CONTACT IS ESTABLISHED PERIOD IF NO CONTACT MADE AFTER 3 HOURS CUTTER SHOULD TRANSMIT MO LONG DASHES ON 375 KCS FOLLOWED BY CALL LETTERS FOR PLANE TO TAKE BEARINGS PERIOD THIS TRANSMISSION SHOULD BE SENT EVERY TEN MINUTES COMMENCING ON THE EVEN HOUR LASTING 4 MINUTES ATTEMPTS WILL BE MADE BY THE PLANE TO CONTACT THE NEAR SHIP AFTER THE MO TRANSMISSION IS FINISHED IN THE MANNER DESCRIBED ABOVE SHIP SHOW SEARCHLGHT AS PLANE APPROACHES ISLAND DURING DARKNESS AND MAKE SMOKE DURING DAYLIGHT UNQUOTE
All of this leaves me confused. On 25 June, Putnam states RDF is 200 to 1400 kHz. On 26 June, Earhart specifies RDF 200 to 1500 and 2400 to 4800. Earhart is not saying 7500. Sounds like she is saying 333 or 545 kHz. What evidence for 7500? Could Earhart state the range in one place but not apply the range at another place? It should at least raise red flags on Itasca.
The 7500 was a last minute radio request at 0758 (Itasca); she responded at 0803, but wanted results set on 3105 kHz.
Were transmissions too short for adequate bearing? Request for long count is to aim DF antenna.
Earhart's test at Darwin was probably at Darwin's ordinary transmitter frequency and not 7500. When was she in Darwin? That would be after Java. Did she evaluate the Darwin test and realize LF or homing 3105 would work? Is the failure that 3105 kHz was not suitable? If she expected Itasca to transmit on 3015 and 6210, then why not home on those signals and forget about 7500? If two receivers, then they could be independent, but with one receiver, the retuning does not make sense.
http://earchives.lib.purdue.edu/cdm/compoundobject/collection/earhart/id/3119/rec/55 page021 page023. AE requested 7.5 MHz! Report notes contradiction. Batteries ran down, but not whole story; Earhart's transmissions were voice (not constant amplitude) and brief (never more than a 8 seconds). page044. Smoke remained concentrated and did not thin out greatly! Where was that picture? Interesting conclusions on page048: Earhart had receiver troubles; ack for 7500 may have been other signals; possible sun line but no position.
Another screwy issue is the gas tanks providing floatation for the plane.
Lockheed had a different conclusion: the plane would sink quickly.
Gross weight is 16,000 pounds (Johnson). Fuel tanks 1150. 9545 Oil tank is 50? 415.
Long and Long also suggest the fuel dump valves would let water in.
When Allen first arrived in Miami and caught up with Amelia at the airport, one of the first things he did was to go over the equipment list to see if there had been any changes since Oakland. he noted one change that he wasn't sure he approved of - the elimination of the marine frequency radio that operated on the 500-kilocycle bandwidth. "Oh," she said, "that was left off when Manning had to drop out of the flight. Both Fred Noonan and I know Morse code but we’re amateurs and probably would never be able to send and receive more than 10 words a minute …The marine frequency radio would have been just that much more dead weight to carry and we decided to leave it in California."
A new, powerful antenna had been installed to maximize the 6210 to be used during the day and the 3105-kilocycle frequency to be used at night. But Pan Am radio technicians decided to replace the antenna with yet another one after she was unable to communicate with either the local broadcasting station or with the Bureau of Air Commerce airways station at the field in a test flight. She planned to broadcast every half hour, fifteen minutes before and after the hour, all the way around the world.
Intended to reach Dakar, Senegal, but they ended up north of their desired position and went to Saint-Louis, Senegal instead. Noonan didn't know why. He passed a note.
Oakland to Honolulu on first attempt:
Luke Field to Howland on first attempt:
Confusion about Quoting C. B. Allen, New York Herald Tribune.
p. 97 left Miami shortly after 6 AM. Tuned WQAM.
p98 6:30 AM Great Reef of the Bahama Banks 7:00 AM Andros Island
p 129 Natal to Saint Louis
p 130
p 217
p 223
There's a huge topic here.
Development. Byrd. P. V. H. Weems. Aircraft navigation in general. Noonan a student of Weems.
There's the octant versus sextant issue. Main source states octant, but there are several pictures that show instruments with a sextant label. Anderson provides a schematic of a bubble sextant, but the scale implies octant.
The accuracy issue and the requirement for averaging.
Then there is what happened to the octant that Manning borrowed from the Navy. I think Long and Long had him return it before the second attempt. Long p. 72: Noonan wanted an octant but had not obtained one. Commander Ragsdale asked North Island Naval Air Station in San Diego if one could be borrowed. Long p. 73: The Secretary of the Navy approved the loan; Manning would have to sign for it. Long p. 103: Manning, who was personally responsible for the loaned bubble octant, retrieved it from the rear of the Luke Field plane crash.
Warner gives a detailed history of the instrument. Many companies are involved: Brandis, Bausch & Lomb, Pioneer, Keuffel & Esser, Bendix. Victor Carbonara, an Italian designer hired by Brandis, designed the Brandis air octant Model 206 in 1925. Warner page 97: "Carbonara realised that air navigators never measured angles greater than 90°, and so could use an octant rather than a sextant." Warner page 104 speculates that Noonan used a Pioneer sextant.
Try a mapframe from mw:Help:Extension:Kartographer. Also mw:Extension:Kartographer.
https://tools.ietf.org/html/rfc7946
https://www.mapbox.com/maki-icons/
There are some problems with lines that cross the antimeridian. On the ordinary display, the western hemisphere lines do not appear; click on the map and they appear. I tried to game it with a longitude of 183, but it does not work. Want to intersect (159.400, -4.517) to (183.383479, 0.807179) with the antemeridian. Near the equator, the world is quasi flat; use linear interpolation to find the latitude:
But that does not do what I want either. The ordinary page display works correctly now (both hemispheres display), but click on the map and the western hemisphere lines disappear. Try using -179.999; does not help. Also, Howland Island disappears. Coordinates are messed up. Using continuous coordinates (Riemann) confounds ordinary display but clicking gives good map. Just shoot me.
GeoJSON suggests cutting geometry at the antemeridian instead of crossing it. See RFC 7946 §3.1.9.
Setting longitude to -179 or 179.
The flight's route is not known. Although Earhart radioed her position from time to time, those reports are not clear.
Although taking a great circle route would minimize the distance traveled and presumably save fuel, there are other considerations.
Long & Long believe they deviated south to avoid some heavy weather. Long supports his theory with a high altitude claim. Long points to a typical 10 percent error in travel.
One radio message copied the position as 157.
The position is too close to Lae to make sense. Long suggests it is the position as of the last fix.
Gary LaPook suggests the position is 150 7'.
LaPook is an attorney specializing in aircraft accidents; he is a celestial navigation instructor and former commercial airline pilot.[12] LaPook believes the crash and sink theory, and did not believe the aluminum found on Nikumaroro belonged to Earhart's Electra. He believes that if Earhart were south of Howland, the weather was clear, Noonan could have taken a fix, and they could have headed for the closer Howland rather than far away Gardner.
Van Asten takes that much further. That Noonan, instead of taking a great circle, went island hopping. Although the Pacific is vast, there are islands and reefs. During daylight, if one sees an island, then one knows his position without doing any difficult celestial navigation. Instead of taking the shortest route, a great circle route, (and fighting the weather), Earhart flew toward landmarks that could be used for navigation. Van Asten believes the Electra was at Nukumanu Islands with a firm navigation fix around dusk. At night, one can take some star fixes to determine position, but one can also look for lights on islands.
On June 30, Earhart received from Nauru:[13]
Need to check the message. The gross longitude error is easy to recognize. The fine error might be convention, but should also be easy to resolve: Nauru is a few miles wide; an error of 0.5 degree is 30 miles. The 5600 feet was an error; the height is only 560 feet (170 m).[14] The beacon height would give Nauru steep sides: a few miles wide and a mile tall. That could have some impact on visibility, but at 10,000 feet, the light would be visible from XXX miles. Clouds could obscure it. The 34 miles is due to the beacon's height and the height of a typical bridge. At 132 (4 times as far), the light intensity would be 1/16th. Binoculars would make up for inverse square loss, but not for scattering loss. Van Asten believes Noonan would have headed for Nauru; if they saw the loom of the industrial lights or the beacon, then they could use it for a line of position. Nauru is enough out of the way that they would not fly all the way to the island. At some point before reaching Nauru, they would turn west to head toward Howland. If they saw the island's light, it would be a bonus.
Van Asten believes Noonan would have chosen a latitude line course. It would be easy to stay on a latitude line by making simple star sights. Van Asten speculates that Noonan would have chosen a latitude line that take them over either Nikunau or Beru Island. Question: looking for lights or visible at morning twilight? Twelve degrees is almost an hour, which is almost 1000 miles. I think the islands are only 500 miles away (three hours of flying). But sunrise and 100 miles from Howland does not fit; that puts the islands two hours back.
There are three versions of twilight: civil, nautical, and astronomical.[15] They occur when the center of the sun is less than 6, 12, or 18 degrees below the horizon.
The calculation looks close. Civil twilight is probably only 20 minutes (5 degrees at 15 degrees per hour). Nautical twilight (when most stars can be seen) would be another 25 minutes. When did O'Kane submerge?
LaPook points out a star fix was unlikely. Earhart reported overcast at 2:45 AM, so a star fix was out.[12]
Van Asten claims the project times match up well (but he can fiddle with the air speed to make them come out right).
Van Asten claims one fix was reported before it could have been made. Need to understand that argument. He offers some explanation about navigators doing their homework (precomputing some values) so they don't have to do the work during the flight.
Locations from reports. Here's a quotation:[16]
Table of locations and projected times.
Radio schedule with Lae would be 18 minutes past the hour. Find reference. AE not heard for 4 hours. Why? AE would be on daytime frequency. When did she announce switch? Lae was upset because she came in loud and clear.
Radio schedule with Itasca was... Itasca picked her up during the night on... Near Howland, AE switched freqs and never heard on new freq. (Long surmises instant of gasoline problems.)
Location | Coordinates | Time | Comments |
---|---|---|---|
Lae Airfield | 06°43′59″S 146°59′45″E / 6.73306°S 146.99583°E | 0000Z | origin |
scheduled radio | 0018Z 10:18AM Lae | not heard | |
scheduled radio | 0118Z 11:18AM Lae | not heard | |
scheduled radio | 0218Z 12:18PM Lae | not heard | |
scheduled radio | 0318Z 1:18PM Lae | not heard | |
Harehare, Shortland Island | 7°08′22″S 155°47′39″E / 7.1393391°S 155.7942539°E | 0418Z 2:18PM Lae | Van Asten guess |
Earhart radio msg | 7°18′S 150°42′E / 7.3°S 150.7°E | 0519Z 3:19PM Lae | Balfour |
Choiseul Island | 7°3′S 157°0′E / 7.050°S 157.000°E | LaPook, Van Asten | |
Earhart radio msg | 4°20′S 159°42′E / 4.33°S 159.7°E | 0718Z 5:18PM Lae | Balfour |
Nukumanu Islands | 4°31′S 159°24′E / 4.517°S 159.400°E | ||
Nauru radio | 1030Z | Nauru radio received "Ship in sight ahead" at 1030Z.[17] | |
USS Ontario | 2°59′S 165°20′E / 2.983°S 165.333°E | 1030Z | TIGHAR 100 knot GC route[18] |
MV Myrtlebank | 2°20′S 167°10′E / 2.333°S 167.167°E | 1030Z | TIGHAR 130 knot GC route |
Nauru | 0°32′S 166°56′E / 0.533°S 166.933°E | Van Asten | |
long run | |||
Tabiteuea | 1°21′S 174°48′E / 1.350°S 174.800°E | Van Asten | |
Beru Island | 1°19′S 175°58′E / 1.317°S 175.967°E | Van Asten | |
Nikunau | 1°21′S 176°27′E / 1.350°S 176.450°E | Van Asten | |
head to Howland | |||
Cross Dateline | July 3 back to July 2 | ||
Earhart radio | "200 miles out" | 6:14AM Itasca | radio log |
Earhart radio | "100 miles out" | 6:45AM Itasca | radio log |
Howland Island | 0°48′26″N 176°36′59″W / 0.807179°N 176.616521°W | destination |
Lae. Harehare. Nikumanu.
Although a great circle route or a rhumb line route is the direct approach, several authors have suggested other routes. Long believes Earhart flew south of the great circle route to avoid some weather. Van Asten believes, "Noonan preferred to fly via as many as possible identifiable landmarks, islands and other contingent aids to navigation, to avoid the hazard of going astray from a long (2,556 mls, 4,113km) great circle route, partially to be flown in adverse weather."[19]
Van Asten believes the route was Lae to Ghaghara, Choisel Island.[20] From there to Nikumanu Islands (where there would still be light to see the islands). From there, the flight would head toward Nauru Island. Nauru has a lighthouse, and van Asten believes the light would be visible from far away. "Nauru (visual range 135 mis from 12,000 ft) when Amelia announced to have the glare of the island's industrial lights and the 560 feet high fixed light in sight."[21] Van Asten believes Earhart would have turned due East before reaching Nauru. By traveling on the 1 degree 23'S meridian, navigation would be simplified. Star fixes could be used to stay on the meridian, and the route would pass over Tabiteuea, Beru, and Nikunau in the morning. From Nikunau, Earhart would head to Howland Island.
Where is Howland Island? Actual. AE thought:[22]
Do they fit in with sighting the ship?
Then Earhart would know her position and be able to head for Howland, a hop of less than 400 miles. Flying with dead reckoning, the error near Howland would be 40 miles.
At sunrise, they would know they were on a sun line, but it would not tell them much about north or south. It also seems that no observation need be done beyond recognizing the break of dawn and noting the time. One could measure the azimuth, but I think a small error there translates to huge distances.
After sunrise, the sun can still be used for further sun lines. The easy method is just advance the sun line by the distance you think the plane has traveled. At 200 miles out, the error might be 20 miles.
Gary LaPook and H. A. C. van Asten disagree. LaPook says don't bother reading Van Asten's EJN articles. They get into a navigation debate. LaPook fails to understand the difference between measuring the upper limb of the sun (first light) and the central position of the sun. The sun subtends about 1 degree, so the error is significant. They also debate the use of instruments. Apparently the sextant always matches the lower limb because that method is the best for accuracy. One just adjusts the sextant to put the lower limb at the horizon. The bubble octant is apparently different. One probably centers a star on the bubble or the bubble on the sun or moon. Consequently, Van Asten seems to have the better arguments. Van Asten did make an error, but that error does not destroy the rest of his argument. There's a debate about the equation of time that needs more investigation. Van Asten seems to wave off altitude considerations. Is taking a shot at 1000 feet the same as at sea level? Van Asten implies altitude is an issue, but thinks it does not affect the calculation significantly?
Plug in numbers at USNO.
Find a text about computing sun lines. Calculate the sun position for plausible locations.
Moon was not up.
At time of sunrise for Howland (1745 UTC), the sun is about 1 degree below the horizon.
NASA says the subtended angle is 0.53 degrees. https://er.jsc.nasa.gov/seh/math30.html See Sunrise; apparent sunrise when 50' below horizon, so the calculation fits. The error above is 4 arcminutes, but I did not use the most accurate coordinates. The sun moves at 360 degrees/24 hour, 15 deg/hour, 15 arcminutes/minute. The 4 arcminute error is about 15 second error. One degree is about 60 nm, so one arcminute is about 1 mile. 1000 mph is 16 mi/min or 0.25 mi / sec; 16 seconds would be a 4 mile error.
The USNO website wants position at sea level.
Itasca log (not the radio log): http://coastguard.dodlive.mil/files/2012/07/Earhart_001.jpg
This document gives 6:14 as 200 miles out, and then 6:45 position as 100 miles out -- an absurd 200 miles/hour. Long is not concerned with that discrepancy.
TIGHAR tries to explain the discrepancy away, but at 200 miles out, sunrise would not have happened yet, so position would be dead reckoning.
The next transmission could have seen sunrise within the last half hour (transmission schedule). Sunrise would be 80 miles off the line. Wait: 1 hour is 1000 miles. One minute is 16 miles. If 200 miles out, then sunrise delayed 7.5 minutes; if 100 miles out, then sunrise delayed 1.25 minutes. Sort of makes sense: sun moving at 1000 mph means 200 miles is only 1/5 of an hour. Would expect report to say 200 miles out; then, 1/2 hour later at 150 mph, to say 125 miles out. Subsequent sun fix may have improved the position estimate by 25 miles. If 100 miles out, then 1/2 hour later (7:15), should be 25 miles out. The 7:45 report would be on the line in either case (except for winds aloft issue that would leave them short). But if we are talking nautical miles, then things are slightly different.
The Coast Guard believed Earhart and Noonan went down northeast of Howland. The sky was clear at Howland, but Earhart reported clouds. That explains why she was flying at 1000 feet: to stay below the clouds. The Coast Guard believed the only clouds were to the northeast. The Itasca was making significant smoke. Earhart could not hear Itasca, so Itasca could not tell them the sky was clear around Howland.
Henry Morgenthau Jr. was a neighbor and friend of Franklin D. Roosevelt. He was Secretary of the Treasury from 1934 to 1945. At that time, United States Coast Guard was under the Treasury Department.
Paul Mantz wrote to first lady Eleanor Roosevelt on 26 April 1938 and asked for her assistance in obtaining Itasca's radio logs and its report about Earhart. Mantz knew that the first lady and Earhart had friends. Eleanor forwarded Mantz' letter to Morgenthau on 10 May 1938. Morgenthau called Eleanor Roosevelt on 13 May 1938, but she was not in, so Morgenthau talked with Eleanor's secretary, Malvina "Tommy" Scheider.[23] Morgenthau stated:[24]
Hello, Tommy. How are you? This letter that Mrs. Roosevelt wrote me about trying to get the report on Amelia Earhart. Now, I've been given a verbal report. If we're going to relase this, it's just going to smear the whole reputation of Amelia Earhart, and my ... Yes, but I mean if we give it to this one man we've got to make it public; we can't let one man see it. And if we ever release the report of the Itasca on Amelia Earhart, any reputation she's got is gone, because — and I'd like to — I'd really like to return this to you.
Now, I know what Navy did, I know what the Itasca did, and I know how Amelia Earhart absolutely disregarded all orders, and if we ever release this thing, goodbye Amelia Earhart's reputation. Now, really — because if we give the access to one, we have to give it to all. And my advice is that — and if the President ever heard that somebody questioned that the Navy hadn't made the proper search, after what those boys went through — I think they searched, as I remember it, 50,000 square miles, and every one of those planes was out, and the boys just burnt themselves out physically and every other way searching for her. And if — I mean I think he'd get terribly angry if somebody — because they just went to the limit, and so did the Coast Guard. And we have the report of all those wireless messages and everything else, what that woman — happened to her the last few minutes. I hope I've just go to never make it public, I mean — O.K. — Well, still if she wants it, I'll tell her — I mean what happened. It isn't a very nice story. Well, yes. There isn't anything additional to something like that. You think up a good one. Thank you.
In addition, there was an undated note on White House stationery sent to Mrs. Roosevelt:[25]
Mr. Morgenthau says that he can't give out any more information that was given to the papers at the time of the search of Amelia Earhart. It seems they have confidential information which would absolutely ruin the reputation of Amelia Earhart and which he will tell you personally at a time when you wish to hear it.
He suggests writing this man [Mantz] and telling him that the President is satisfied from his information , and you are too, that everything possible was done.
On 14 May, Eleanor Roosevelt sent Mantz a letter that followed that suggestion:
I have made inquiries about the search which was made for Amelia Earhart and both the President and I are satisfied that the information which we have received that everything possible was done. We are sure that a very thorough search was made.
On 5 July, Morgenthau told Eleanor Roosevelt that the government would release the Itasca radio logs to Mantz. On 21 July, Coast Guard Rear Admiral R. R. Wesche (*** check USCG Commandant) sent Mantz the logs.[26]
Colorado arrived a week after and searched with its planes.
Lexington arrived on the morning of 13 July 1937 and launched 60 planes for the search. Stopped search 5 days later (28 July). The group had searched 262,281 square miles.[28]
Coast Guard Captain Thompson wrote its commandant Rear Admiral Russell R. Waesche on 18 October 1937, "As regards the search, I was very sanguine up to the time the Lockheed people stated that her pane could not possibly be landed without crashing and sinking."[29]
Gas on board is an issue for the Lae, New Guinea, to Howland Island leg of 2,556 miles (2,221 nmi; 4,113 km).
There are three different fuel amounts that are claimed. News reports shortly after the disappearance say 950 gallons, Earhart told local authorities about 1000 gallons, and Chater says about 1100 gallons.
The airplane capacity is 1150 gallons. A 950 gallon figure is had by filling all but the two front wing tanks (97 gallons each). A 1000 gallon figure is filling all but the two front wing tanks and including 50 gallons of 107? octane full that remained in one of the front wing tanks. The 1100 gallon figure is filling all the tanks except for one front wing tank that had 50 gallons of 107 octane fuel. A consistent figure is 1000 gallons. In addition, Earhart did not want extra weight (an extra 100 gallons weighs 600 pounds), and Putnam stated that Earhart was practicing take offs with 1000 gallons. (The map in Earhart's book gives the figure as ????.)
Long also accounts for fuel expansion.
How is the 900 gallon Hawaii figure reached? Don't fill one of the 149 gallon tanks and don't fill one of the 100 gallon tanks.
On the first world flight attempt, Earhart used 900 gallons for the Oakland to Honolulu leg of 2,400 miles (2,100 nmi; 3,900 km). The estimate was that gave a safety margin of 40%. When Earhart took off from Luke Airfield on the Honolulu to Howland leg of 1,900 miles (1,700 nmi; 3,100 km), she also had 900 gallons on board (and two other passengers).
If she didn't put any gasoline in the front wing tanks, she'd be about 200 gallons below the max, or 951. Put 50 gallons in one wing tank and don't fill the 70 gallon rear tank?
How much for Natal, Brazil, to St. Louis, Senegal 1,727 miles (1,501 nmi; 2,779 km)?
Earhart told officials at Lae that she would probably carry 1000 gallons.[30]
Longworth quotes Nesbit's articles in January and February 1989[31]
Black states (also believes crash and sink):[32]
18 55-gallon drums is 990 gallons. She could assume there would be fuel left in her tanks when she landed.
Chater says she had 1100 gallons, but that is based on his belief that all tanks being full except for a half-full wing tank with 100 octane fuel.
TIGHAR states (diagram also shows position of transmitter):[33]
Total capacity (using 102 gallon wing tanks): 1151
Pretty clear that one front wing tank (capacity 81+16 gallons) was half full, then there would be about 1100 gallons. If the other front wing tank were also empty, the plane would have about 1000 gallons.
Leaving 100 gallons at Lae saves 600 pounds.
Tne Navy considered the fuel consumption of Earhart's plane. Economical airspeed was 130 knots (150 mph) and burning 45.8 gallons per hour and 24 hours of endurance. At 140 knots (161 mph) the plane would burn 53 gallons per hour and have 20.75 hours of endurance.[34]
Calculation checks: economical range is 3602.6 miles; fast range is 3341.5
Long & Long (1999, p. 233) interpret Kelly Johnson's fuel consumption information, determined the average airspeed was 160.5 mph, and the endurance to be a little over 20.5 hours.
Long & Long (1999, p. 232) buttresses the 1/2 hour of fuel statement. Thompson and seven other people confirmed the statement, and there were three contemporary written records: the two radio logs and some handwritten notes by United Press correspondent Howard Hanzlik.
Is there a simple way of doing this? Or is that what Elgen Long did? Effective distance is increased by headwind. 25 knots * 20 hours is another 500 nm. (nm v statute m issue.) At 140 knots, that's another 3.6 hours at 53 gph or 189.3 gallons. The flight leg of 2556 miles at 150 mph would take 17 hours and 780.4 gallons.
Kelly Johnson predicted a 40% reserve on the Oakland to Hawaii flight, but there was a tailwind.
Earhart predicted the flight would be 18 hours.[8]; Lae sent Itasca this message for Richard Black:
That would mean hitting the sun line about dawn? Check: Lae 10 AM is 0000GMT advance 18 is 1800GMT subtract 11.5 is 0630 Itasca. Close to sunrise at 6:15 Itasca time. The 18 hour flight time would mean 142 miles per hour (123 kn).
David K. Bowman's website.
General coverage of theories
Upper limb issue
Bendix / Hooven guy was among the first to raise the possibility. He noticed the radiolocation of the post loss transmissions. He discarded the theory after the flyover didn't find a plane.
If one believes the GIH, then one must account for the fuel. If the plane were flown at economical speed, then it would have 24 hours of endurance. Kelly Johnson recommended flying at economical speed even when confronting a 25 (mph or knot) headwind. Without headwind, Earhart could make the flight in 18 hours and have 6 hours of fuel remaining. With a 25 knot headwind, Earhart flew another 500 nmi. That could add 4 hours to the trip time, but it only took another 2 hours. Therefore, Earhart flew 10 percent faster than economical speed. Also the detour to avoid weather. But economical speed also varies. Look at Johnson's report. Earhart radio messages imply Johnson's recommendations were not followed. High altitude early. High speed early. Density altitude; speed reports. In particular, one must counter Earhart's statement about 1/2 hour of fuel remaining. TIGHAR suggests the 1/2 hour of fuel remark was until they had to start using the reserve. TIGHAR also points out the plane was still transmitting an hour later; other point out a shift to maximum endurance would increase time aloft. TIGHAR also argues that Earhart did not waste time searching for Howland but immediately went for Baker and then the Phoenix group. Witnesses comment about Earhart's stress being consistent with running out of fuel.
If one believes the transmission on XXX came from Earhart on Gardner Island, then one must explain the plane staying on the island for 6 days and then disappearing in the time from last transmission to flyover. Bendix / Hooven guy discarded the theory for that reason; he later adopted the Japanese took the plane.
If one adopts the Japanese took the plane, then one must explain why. The Japanese could be heroes for finding her. If the plane was in the Phoenix Islands, then it would be an unlikely spying mission. If Earhart had film, then the Japanese could confiscate the film and still be heroes. Earhart's path went over.... Also, daytime photography would be observed; nighttime photography may not show much. Why would the Japanese be around Gardner a week after the crash while the US was conducting an intensive search?
If one believes the (non-native) skeleton is Earhart, then one must account for the failure to remark about the presence of surgical holes in the skull. Earhart had two holes made to relieve sinus issues.
If one believes the Lockheed made it Gardner, then one must explain why Howland was not seen. AE was flying at 1000 feet presumably due to cloud cover, but the sky was clear at Howland. If AE had come south from the known clouds, the sky would have cleared, she would have climbed to get a better horizon, and then Howland should have been seen. Consider Itasca making smoke and image shown in one AE book.
Was the Navy was right? Navigation put her too far north, so she was just circling there? If she had headed to the Phoenix Islands, she would have seen Howland (smoke) or Baker (low to the sea).
Any event seems to say a significant navigation error. Too far north and stuck in the clouds, or too far west so cannot see island even in clear weather.
Other people are possible castaways. The shipwreck lost 11 crew, but only four bodies were recovered.
If one believes the shoe is hers, then one must account for the shoe being size 9 while Earhart wore size 6.
If one believes the aluminum piece is from the Electra, one must account for the non-1937 aluminum grade and that the hole pattern fits a WW II plane.
Wasn't there a WW II plane crash? TIGHAR had said that a plane was visible on the island.
Fuel and flying time is an issue for the hypothesis. Earhart left Lae at midnight GMT. The "WE MUST BE ON YOU" transmission was at 7:42 am Itasca time, which would be 20:12 GMT — a little over 20 hours into the flight. If the plane followed Kelly Johnson's schedule with 1100 gallons of gasoline, then the Electra would still have four hours of fuel on board. That would be enough to make it to Gardner Island, which was a little over 2 hours away at 150 knots. However, Earhart's message also stated the plane was running low on gas and that there was only half an hour left. One explanation is that the Electra had only a half hour before it would start using its reserve supply. Certainly the plane was still in the sky an hour later when the last known transmission at 8:43 am was heard.
Speed over ground would be 127.8 miles per hour (111.1 kn; 205.7 km/h). Expected flying time at 150 mph would be 17 hours. Expected flying time at 150 kn would be 14.8 hours. Suggests headwind of 40 kn. Knew it was 27 knots at one point. One alternative is to step on the gas.
Did they want to be over Howland before dawn at 6:15? Leaving at midnight, flying for 16 hours, would put them precisely at dawn. Half an hour before dawn would be 75 nmi away and right at the horizon. They believed they were 200 nmi out at dawn.
Flight from Oakland to Hawaii was 16 hours. Speed 150 miles per hour (130 kn; 240 km/h) (distance is 2400 miles).
But the castaway theory is full of holes. Nikumaroro lies 350 nautical miles south of Howland, and Earhart herself reported that she was running out of fuel near the island. Those radio transmissions supposedly picked up by random people thousands of miles away? None have been verified as coming from Earhart. Navy search planes flew over Nikumaroro a week after her disappearance and saw nothing related to the aviator: not a human, not an airplane or the debris of one, not a smoke signal, not an SOS written with palm fronds.Also doubts Jantz and quotes Jantz.
Torpedo depth control. Robert Whitehead in Fiume had trouble with depth control until it developed the "balance chamber" with pendulum. The balance chamber had water pressure push against a disk (cylinder?) that was balanced by a spring. "The inclusion of a pendulum stabilized the mechanism's feedback loop." Newpower (2006, p. 12) citing Edwyn Gray, The Devil's Device: Robert Whitehead and the History of the Torpedo (Annapolis MD, US Naval Institute, 1991 p33. This was sometime around 1870.
Whitehead had an improved exploder around 1877. Used a firing pin, but also had an arming mechanism driving by an impeller. (Newpower 2006, p. 15)
The US Navy tried developing its own torpedoes, but did not have much success. In 1891 E. W. Bliss Company signed a licensing deal with Whitehead to produce its Mark 1 torpedo. The US Navy started buying torpedoes from Bliss for $2000 each. (Newpower 2006, p. 15) It effectively ended the use of the Howell torpedo. (Newpower 2006, p. 16)
Around 1904, E. W. Bliss had its own torpedo that matched the Whitehead Mk. 5: the Bliss-Leavitt Mk. 1. That design had problems with roll stability, but Navy Lt. Gregory Davidson proposed using counter-rotating propellers. The proposal was adopted. (Newpower 2006, p. 18)
Navy decides to build its own torpedo factory. Newport breaks ground in 1907 and starts production in 1908. (Newpower 2006, p. 19) (But this has torpedo station producing Whitehead Mk. 5s!) 1912 saw the Bliss-Leavitt Mk. 7 as the last torpedo ordered by the Navy. Mk. 7 was the main torpedo of WWI.
p20 in 1911, Navy starts making its own contact exploder. Results in Mk 3 exploder in 1915. Also anti-circular run device.
Start with Navy buying torpedoes from other vendors such as Whitehead and Bliss. Then it learns all it could from Bliss and starts doing its own production and development.(somewhere in Iron Men). That should be the Mark 10 production. TPS does not have the development or manufacturing expertise. NIH takes over.
The US Mark 10 is a reasonable torpedo, and the S-boats uses them effectively during WW II. (Newpower 2006, pp. 116–117)
Construction/design is by assemblies.
Exploder comes first.
Mark 3 exploder. Did it have counter-mining immunity? There was a friction issue that was solved with a stronger spring.
Where was Einstein and the crush head?
Worlds navies are recognizing that armor/blisters are becoming more effective against torpedoes. There are different responses.
U.S. believes TNT is good enough; later switch to Torpex; Wikipedia article says it was used starting in late 1942.
TPS also increased packing of warhead. US decides influence exploder is the trick (avoid the armor). Compare Germany (who developed MI and later copied British torpedo), UK (which did not oppose a military fleet (Britannia ruled the waves), so a contact exploder was good enough), and Japan (massive amounts of TNT).
G.B. doesn't want to redesign torpedoes so it looks for more powerful explosive; finds it in RDX. (MacDonald and Mack Partnership 1984)[page needed] G.B. also looks at influence exploder but decides it is unreliable and just uses contact exploder. G.B. has reliable contact exploder and reasonable weapon, but limited targets. Germany does not have significant shipping or risk much of its fleet. Compare Hood, Ark Royal, and Bismarck.
Germany develops new steam and electric torpedo and uses magnetic influence exploder. Doenitz goes nuts about depth control and enabling/disabling the MI exploder. Contact exploder is unreliable, too. Norway campaign is a disaster. German steals G.B.'s contact exploder.
Japan goes the more explosive is better route. Type 91 airborne torpedoes. (Newpower 2006, p. 113). Type 93 torpedo Long Lance has 490 kilograms (1,080 lb) and Type 95 torpedo has 893 pounds (405 kg). (Newpower 2006, p. 122) but another says 405kg model 1, 550kg model 2! Big explosive rather than the MI feature. Long Lance recognizes need for depth control. Japanese torpedoes are devastating: Pearl Harbor and Guadalcanal, but Japan does not deploy her submarines effectively;(Newpower 2006, p. 113) combat air cover does reasonable job of keeping torpedo bombers away.
Lockwood likes electric torpedoes.
Some German electrics land on US shores in 1942, TPS wants nothing to do with them. Westinghouse copies them (with GB contact exploder?).
In the summer of 1926, Lieutenant Ralph Waldo Christie became involved in the exploder's design.[1][2]
Should explain the nomenclature. The Mark 14 is the torpedo minus the war head, but the term is often used to refer to the whole torpedo.[citation needed] Internal to the Mark 14 were several assemblies. Cite to the pendulum depth keeping mechanism; it should be used on several other torpedo designs.
There's a war head that attaches onto the Mark 14. The warhead includes the explosive charge, but an exploder assembly attaches to the warhead.
The Mark 6 exploder bolts into the war head. There's a magnetic sense rod that then slips through the induction coil. The Mark 6 has an impeller wheel that is pushed by the flow of water. If drives a generator to supply power to the electronics; it also hoists the detonator into the war heads charge. If torpedo ran hot in the tube, no impeller motion, no arming of the torpedo, no charging of the capacitor.
Comment about the independence and interchangeability of the various Marks. Blessing and a curse. Balance issues.
In lieu of the warhead-exploder combination, there was an exercise head.
Both the warhead and the exercise head had a lead ballast on the bottom. Function as a keel.
Page 15, OP 635, 24 March 1945 says
Early torpedo: Mark 10 torpedo (this warhead was 500 pounds). Was warhead interchangeable?
Family of torpedoes:
Depth setting problem
Who described depth sensor being moved inboard to cooling flow?
Air flask supplies oxygen. Methanol for fuel. Drives turbine.
Two speeds.
Section floods with water to cool engine.
Gyro and depth are one subassembly, the Mark 12-3.[3] The same assembly is used on the Mark 18 torpedo.[4]
Heading control. Gyro.
Depth control. Depth sensor and pendulum. Pendulum needed for stability. Consider way too deep, torpedo would just loop at depth. With pendulum, ailerons go neutral at reasonable attack angle.
Manual, pages 99–108. Uhlan principle. Page 98:
The pendulum has another effect. When the torpedo is initially accelerated, the pendulum swings aft. Consequently, the torpedo thinks it is pitched up. The depth rudder will try to correct for that by ordering submerge. Consequently, the torpedoes go deep at first.[5]
Explain doubled instrument error. BuOrd "roll recorder".
Multiple faults: http://www.ibiblio.org/hyperwar/USN/Admin-Hist/BuOrd/BuOrd-6.html (Chapter 6)
Believe same mechanism used in the Mark 10 and possibly the Mark 18. There needs to be some adjustment for speed if the sensing positions are not the same. Can I find the sensing positions in diagrams. Or are there references that discuss the repositioning of some ports.
Is Bernoulli's principle explanation flawed?
Calculate Reynold's number (stolen from that article).
The Reynolds number is defined as[6]
where:
From Viscosity article:
Looks like 12 million.
According to late manual (pages 13–14)
Generally, I think the A suffix is improved depth control (port moved from side to the interior).
Manual, page 15:
Manual, page 16:
As I understand it, the original design used a charge was 500 pounds.
Some named person increased charge to 640 pounds or so. How that was done is debatable. One claim is the length of the warhead was slightly increased (could that be a reason the torpedoes did not fit in an S-boat?), but the exercise head was not changed.
Shireman states:
That is not the complete answer. 500 + 115 = 615 pounds. Original warhead was TNT (1.654 g/cc), but Torpex (1.72 g/cc[7]) is better (but appeared later). There are some density issues, too. Shireman's comments are a little suspect. One cannot change the density of TNT. There are comments (I think in the Mark 14 manual) about placing extra TNT in a space that was previously void. See Manual, figure 4, page 15.
If the torpedo became nose heavy (but it was balanced, wasn't it?), then some up aileron is needed, so the torpedo runs deeper.
The torpedo is 20 feet 6 inches. Why the 6 inches? Let it all be an addition to the warhead, so it would create increased volume. Assume the increased volume is a 20 inch diameter cylinder (10 inch radius):
That corresponds to 1885/231=8.16 gallons (8.159980918415). A gallon of water weighs 8.33 lbs, so 67.9728 lbs of water. TNT has a density of 1.654, so 112.4270112 pounds (116.913216 pounds of Torpex), close to the 115 pounds in some references.
Could expect the cavity to hold 500 pounds of TNT. That would be about 520 pounds of Torpex. Add 115 pounds for extra length to get 635. A 31 lb topping charge would give 666 pounds.
Overall length of war or exercise head is 47.28 inches (Mk 16). But summation of lengths gives 20.5 ft, so Mark 16 may refer to later, or there may have been a short warhead (for Mark 10?).
Mark 14-3A torpedo used (manual, page 275). (16, 16-2, 16-3, 30, 30-2, 30-3 added to table.)
Mark | type | 14-3A | total weight | empty weight | lead ballast | liquid ballast/payload |
---|---|---|---|---|---|---|
16 | warhead | N | ??? | ??? | ??? | ??? When loaded with Torpex, there is a cavity to maintain trim (manual pg 16) |
16-1 | warhead | Y | 1053 | 264 | ??? | 666 lbs Torpex (said 66.6 lbs) TNT topping charge 999info says pre/early war 507 lbs TNT; later 643 lbs Torpex |
16-2 | warhead | N | ??? | ??? | ??? | ??? |
16-3 | warhead | N | ??? | ??? | ??? | ??? |
16-4 | warhead | Y | 1057 | 264 | ??? | 600 lbs Torpex TNT topping charge 999info says 666 lbs Torpex |
30 | exercise | N | ??? | ??? | ??? | ??? |
30-1 | exercise | Y | 843 | 429 | 287 | 414.5 (water); end buoyancy 210 |
30-2 | war exercise | N? | ??? | ??? | ??? | ??? |
30-3 | war exercise | N? | ??? | ??? | ??? | ??? |
30-4 | exercise | Y | 1053 | 503 | 367 | 550 (sp. gr. 1.35) (414.5*1.35=559.575) Seems to recognized that 30-1 head is 210 pounds light; 135.5 lbs liquid ballast added; 74 lbs lead ballast; end buoyancy 136 |
For war head, total weight may not include exploder. For simple exercise heads, total weight is empty weight plus liquid ballast.
Lead ballast (pull around) exists in warhead but warhead lead ballast not specified in appendix.[8]
999info is http://www.999info.net/Grunion/USN%20Weapons.pdf
There were several Marks. See appendix in the manual. These different heads probably give a clue about the warhead size changes.
Manual, page 21:
There is a war exercise head that is used for .[9]
The Mark 30-3 is such a head (it is called out in Figure 16). Figure 17 suggests that the Mark 30-2 is also a war exercise head (marker bomb used in 30-2 would require a functioning exploder). The 30-4 is probably not a war exercise head because it is not of phosphor bronze. The only purpose of demonstrating the warhead would be to show of the MI feature. Not clear a torpedo would survive a contact on a steel hull.
O'Kane may describe exercise runs where torpedo is set to run underneath the target.
BuOrd pg 98:
Somewhere in Newpower (2006), at the tail end of the torpedo crisis, Blandy talks about a new exercise head.
The British also needed to upgrade their torpedos, and they experimented with a magnetic influence exploder. The test results were unsatisfactory, so the resulting torpedo, the Mark VIII, had a simple but reliable contact exploder. There was no magnetic influence feature.[10]
The British were aware that capital ships were being armored against torpedos, so the explosive power of the World War I torpedo would no longer be adequate. That presented a design challenge: either increase the amount of explosive or find an explosive with more energy. Increasing the amount of explosive implied that existing torpedo tubes might be inadequate. Consequently, the British developed the more energetic explosive RDX.[11]
The British had a reliable torpedo when it entered the World War II. Perhaps the only disadvantage the British had was the limited number of targets for the torpedo. Brittain is an island nation that depends on shipping, so there were many targets for German submarines. German submarines attacked both British capital ships and merchant shipping. Germany's resources did not depend so much on shipping. Brittain had some success against the German invasion of Norway.[12] (There should be some comment about Brittain v. Japan in the Far East.)
The U.S. was behind on more powerful explosives. The U.S. military had decided that TNT was all the explosive it needed. The U.S. bet on the magnetic influence feature avoiding the armor, but then apparently retreated somewhat. There are stories of the U.S. adding another 116 pounds of explosive to the warhead (something that made the torpedo nose heavy and resulted in deeper running).[13] The British disclosed the existence of RDX to the U.S., and the U.S. started manufacturing it in 1940 (???). The availability of RDX led to the use of Torpex in U.S. torpedos.[14]
The Japanese developed several successful torpedoes. For the problem of ship armament, the Japanese just used more explosive — a lot of it. Cite warhead sizes. Newpower (2006) describes damage to U.S. Heavy Cruiser. Cruiser damaged but made it to port. Sent home for further repairs, but sinks on the way due to unrecognized structure damage caused by the torpedo.
The German torpedo of the time also had depth control problems due to using a light exercise head; the German torpedo ran 6.5 feet (2.0 m) deep.[15] The German torpedo designers knew the torpedo ran deep, but they ignored the problem because the magnetic influence exploder would make up for it. Some German officers were confined because they had let the problem persist.[16]
The German torpedo had problems with it magnetic influence feature; there were prematures.[15] The Norway campaign was an eye opener. There was some belief that ore deposits were providing fields that triggered the feature. Doenitz went back and forth about using the feature.[17] Ultimately, it was judged not reliable.
The German torpedo also had problems with its "clumsy" contact exploder.[15] The Germans captured some British Mark VIII torpedoes, which had a reliable contact exploder; the Germans simply copied its contact exploder.[18] Interestingly, some unexploded German electric torpedoes were later recovered from some soft, sandy, East Coast beaches. Westinghouse then copied the German electric torpedo rather than designing an entirely new one. The result was the Mark 18 torpedo, and it essentially used the British Mark VIII contact exploder design.[19]
Although the Germans had essentially the same problems with their torpedos, the Germans resolved those problems in about six months.[20] It would take the Americans a couple of years to work the bugs out of the Mark 14 torpedo.[21]
The manual indicates that the Mark 30-1 and 30-4 are the same (long) dimensions, and some reverse engineering can be done. The Mark 30-1 is lighter than the Mark 16-4 warhead by 210 pounds, so it would not provide an accurate firing simulation. Making the exercise head heavier by 210 pounds would improve the simulation, but in order to recover the torpedo at the end of the run, it must have significant positive buoyancy. At end of run, the Mark 30-1 only has 210 pounds of buoyancy, so adding 210 pounds of lead ballast would not allow the torpedo to be recovered. From the tables in the appendix, the density of the liquid was increased to 1.35 (adding another 135.5 pounds) (volume is 100 pints? previous volume was 70 pints?). The lead ballast was increased by 74 pounds. (Pints of lead = 6.5255731922399.)
There is some evidence that Lockwood's fishnet test used a high density calcium chloride solution to match the warhead weight.[22][23] The test was probably done with a Mark 30 (short), but it is not clear only calcium chloride was used to increase the ballast.
Shireman:
At BuOrd, Commander James King tried Lockwood's test and confirmed Lockwood's results.[24]
Shireman:
The above does not seem quite right; difficult to change density. Adding 6 inches to war head does make sense.
Manual is opaque about the preparation of the 1.35 specific gravity solution. Manual, page 121:
Mark 18 says use 1.355 specific gravity solution in specifications (Mark 18 page 4: "620 pounds of calcium chloride solution having a sp. gr. of 1.355"), but directions for exercise say fill with fresh water (Mark 18 page 153: "Fill head with fresh water").
There are some intriguing possibilities here. The initial depth tests may have been accurate with a Mark 16 warhead. The Mark 30-1 exercise head may be a lengthened (and heavier) Mark 30 head, and it may have shown the 3-foot depth error (manual corrections replace 10 feet with 13 feet). It may have required the extra 210 pounds of ballast to show the 10 foot depth error that Lockwood found. (Attempt to explain why BuOrd would claim 3 feet deep after Lockwood's fishnet test showed 13 feet.)
The Washington state torpedo station used sight depth tests first; later it used fishnet tests.
File:Mark XIV torpedo warhead cutaway.JPG
German magnetic influence mine. Germans didn't use it on a torpedo.
Contact exploder. The safety elevator made the firing pin motion perpendicular to travel. Would work at high angles or low speeds. Early Mark 6 had a complicated inertial ring mechanism; later was a simpler ball switch.[25]
It is not clear that the inertial ring mechanism was flawed. The problem that Lockwood uncovered was with the firing pin mechanism undergoing severe deceleration. The same firing pin was used for the magnetic influence trigger, but the torpedo is not decelerating at instant of triggering. The Mark 6 exploder still used an electronic circuit to trigger the detonation (but it may have bypassed the thyraton). It would be interesting to find out if the thyraton failed under high deceleration.
Shireman points out that BuOrd had already seen this problem in the 1930s. BuOrd tests found the firing pin was unreliable, but it solved the problem by using a more powerful spring to overcome the forces. Unfortunately, the spring worked at 30 knots but not at 46 knots:
It is not clear if BuOrd did the tests at full speed. When did the Mark 14 have its high speed? Or were the tests done for lower speed torpedoes?
After the torpedo is launched, sea water travels through an open channel on the bottom of the torpedo. The water hits an impeller that turns a shaft connected to a generator. The generator will supply power to the exploder. There is a worm on that shaft, and it engages the worm wheel that spins a vertical shaft. There is another worm on that shaft (not in the schematic but on the other picture), and it turns the delay device gear. Initially (A), a switch touching that gear grounds the generator field winding and prevents the generator from producing any power. (That should make the the generator shaft difficult to turn.) Notice that some teeth are missing on the gear at the 11 o'clock position; there's also a hole in the gear at the 3 o'clock position. This gear will make almost a complete turn and reach (B). At that point, the hole is underneath the switch and causes the switch to stop shorting the field winding. The generator starts producing electricity to power the exploder. At the same time, the missing teeth are now at 9 o'clock, so the worm no longer engages the delay gear; the delay gear stops moving.[26]
The vertical shaft continues to turn and has been doing other work. Through a gear train, it causes the arming gear to rotate. That causes the safety chamber to rotate and has the detonator holder rise out of the safety chamber. When the detonator is inside the safety chamber, it cannot detonate the tetryl booster and keeps the weapon safe. Once it rises out of the safety chamber, the booster can be detonated.
With the early Mark 6, something else is also going on. Initially, the threads on the arming screw prevent the screw from moving up and keeps the trigger plate from rising. As the arming gear rotates, it also raises the arming screw. Raising the arming screw further compresses the firing spring. Eventually, the arming screw runs out of screw thread, so it stops just above the arming gear but below the firing pin. The arming screw still retains the ball(s) that holds the firing pin in place. When the torpedo hits something, the firmly attached exploder body suddenly decelerates, but the firing ring has inertia and keeps moving. The trigger plate moves upward and raises the arming screw high enough so the ball(s) leaves the groove in the firing pin; the firing pin is then driven by the firing spring and strikes the detonator.
In the early Mark 6, detonation is even more convoluted for the magnetic influence feature. A thyratron tube fires and energizes the solenoid which moves the armature. The solenoid does not have enough energy to cause detonation, so it just moves the pawl so it will engage the spinning ratchet that is also on the vertical shaft. The ratchet moves the pawl which moves the pawl arm. The pawl arm then moves the firing ring which initiates the same sequence as above.[27]
The later Mark 6 did not use a mechanical detonator. It still used the arming gear to raise the charge out of the safety chamber and the delay device on the generator, but a simpler electric detonator was used.
Many belived in MI, and the feature did work at least once in awhile. Wahoo down the throat shot.[28]
Torpedo scarcity drove desire to use MI. There's some strangeness here because most of U.S. submarine attacks were against merchantment that had no armor. Tang was sinking merchantment with single torpedos late in the war, so it was not required to do the under the keel shot.
Magnetic influence exploder. Disable by removing core rod. Does anyone discuss degaussing of ships and impact on torpedo? Newpower 2006 does.
Interpreting the bubble track.
Running deep would take torpedo out of the influence of the ship.
Who described there were two tests and one failed. That was a torpedo article in several parts. I need to find them. (all at archive.org?)
If the MI feature were designed so that the war head would detonate when it was 10 feet under the keel, then it could also detonate when it is 10 feet away from the hull. That's probably why the MI feature needed to be disabled to avoid premature detonation. If the ship were not degaussed, then prematures might be more likely.
Higher torpedo speed would make the MI more sensitive. The earth's magnetic field is static; a ship's magnetic field would also be static. The induction coil senses a change in the field. If the relative speed to the target is 50% faster, then the signal is 50% larger. Was the MI designed for low-speed torpedoes and then used on high speed torpedoes?
BuOrd 100–101:
Telling comment on page 101:
BuOrd page 102 about prematures:
The original contact feature was an inertial ring. (There's a reference for that.) At contact, the ring was supposed to slip out of place and release the firing pin. A spring would drive the firing pin into a primer, and that primer would initiate the tetryl explosive booster charge that sat on top of the detonator housing. (Loose fitting is described in the manual procedures for assembling the exploder to the war head.)
A safety feature is detonator sat in a safety housing, and the primer could initiate the booster charge without setting off the main charge in the war head. When the torpedo started its run, an impeller would spin and raise the detonator (and the booster charge) into a well of the warhead. The more confined space would allow enough of a shock to detonate the warhead's main charge.
A further safety feature may have been the raising of the detonator mechanism also cocked the firing pin spring.
The magnetic influence feature also needed to detonate the main charge. That was accomplished with an induction coil sensing a change in the magnetic field and producing a voltage. That voltage ultimately triggered a thyratron that energized a solenoid. The solenoid would trip the firing pin.
The problem with the setup was the firing pin moved perpendicular to the torpedo's travel. At impact, there would be high deceleration forces that would interfere with the firing pin movement (and even distort the pin). Consequently, the BuOrd replaced the spring, firing pin, and primer with an electrical detonator (blasting cap). For contact firing, a new sensor was needed to fire the electrical detonator, and that sensor was the ball switch.
Comments about leaving many other things the same.
Air flask destroyed on impact, spray of water against the ship, but not high order detonation.
There were some crazies here. Perhaps Einstein consulted on the Mark 3 exploder problem. Failure at higher torpedo speeds. Result was stiffer spring.
Little is said about the Mark 5 exploder (Mark 6 without the MI that was issued to most submarines).
Many claimed duds. Daspit final straw. Fire into HI cliffs. Third was a dud. then the drop tests.
Lockwood ordered some famous tests, but the some details are not clear.
The fishnet test. At this point, it's not clear that the exercise heads were updated. One source suggests the use of a denser solution, but it is not clear who is responsible for that. BuOrd said it ran its own tests or calculations, and admitted 3 feet. If BuOrd ran fishnet tests, then why didn't it come up with 11 feet as did Lockwood? Did BuOrd use a light warhead with fishnet? Or did it use a compromised depth gauge? Why would BuOrd come down on Lockwood so hard? The fishnet experiment is pretty basic. Fishnet could deflect before ripping, but 8 feet? Any information about how the fishnet was anchored above and below?
Bluff firing. Real warheads, so hard to dispute conditions. The some wag comment about going under the island.[29]
The drop tests seem most poorly reported. Filling the warhead with concrete just seems silly, but that is reported in Newpower (2006) by a guy who was there. Dropping entire torpedoes just to test the warhead also seems silly; why destroy a expensive Mark 14? Dropping a head full of explosive seems a bit dangerous (what happens when it is a dud?). The test must be sure the primer actually fires. Operation Pacific shows dropping just the warhead, but the exploder shape is wrong. Also, the real test would require arming the torpedo just before he drop. If he used real warheads, then removing the explosive would be troublesome. If he had a supply of right exercise heads, then he could use those (the simplest route). If the explosive were absent, then what was used as a filling? Concrete seems more trouble that it is worth. An exercise head is easily filled with liquid. An emptied warhead could be filled with a small amount of sand (which is much denser than the explosive); there is no reason to have the inert material take a set. The concrete would make sense as a replacement for Mark 14; much cheaper and reusable.
Result: 7 out of 10 failed at perpendicular.
Dropping just the warhead without an equivalent Mark 14 mass behind it could show a higher acceleration because the nose does not crush as much.
The drop from 90 ft. d = 90 feet = 0.5 a t2. Acceleration of gravity (a) is 32 ft/sec/sec, so t = sqrt(90/16) = 2.3. Thus v = 2.3 * 32 = 76 feet per second (45 kn).
About the same time as the drop test, there was a chicken wire test. Some skippers believed the torpedo was still running deep. Lockwood's tests showed the depth was about right.[30]
The explanations here are confused and varied. The prematures hid the failures of the contact exploder.
Arming distance. If the torpedo armed before settling down on a course, then the MI might fire.[31]
Gaskets leak. A real shock. If torpedo was immersed for 15 minutes, it got water logged. There were many discussions about gaskets and whether sub personnel did the seals right. Anecdotal story about Gallatin believing impeller packing was not tight enough. Mechanism is salt water causes a short that triggers exploder as soon as it arms.
Infamous BuOrd draft letter. Degaussed ships may not trigger MI. Ungaussed ships may trigger MI far away. Latitude makes a difference. The close aboard detonations.
Mark 14-3A Torpedo and its MK 6 Exploder
I have ten torpedo related questions for you.........
Sommerfeld 1908
was invoked but never defined (see the help page).Looking for details. History of Mark 3 exploder? Christie was only a lieutenant when working on Mark 6; who was in charge? Mark 10 evolution.
Where is the targeting advice letter? Newpower, Chapter 8, Note 80.
Newpower also has Blandy's early claims to performance and subsequent performance. Subsequent performance showed that shallow running torpedoes would prematurely explode about 9 percent of the time.
Also, the test firing data of the Mark 10 torpedo with the MI head involving USS Indianapolis (CA-35) were apparently lost. Southern latitudes were not good for the MI. Newpower ch 8.
xxx
Somewhere there were details about the May 1940 introduction of Kennbuch procedures and then a fast break. I think it was Tony Sales' pages, and I probably put it in the talk page timeline.
There were several pinches. Sales' http://www.codesandciphers.org.uk/virtualbp/navenigma/navenig6.htm
Seizing the Enigma has the milchcow attack that went awry.
Tony Sale's site has lots of info, http://www.codesandciphers.org.uk/enigma/index.htm
M4 project, http://users.telenet.be/d.rijmenants/en/m4project.htm
Enigma Breaking Project has details, messages, German code forms, http://www.enigma.hoerenberg.com/index.php?cat=Welcome
Steven Hosgood gives details about Banburismus, http://stoneship.org.uk/~steve/banburismus.html
There are several issues to understand here.
Bombe menus.
The doubled indicator procedure could find a stop. If it were right, then the wheel order and start position were known, but the ring position and steckers were not.
If someone started decrypting the message, it would turn to gibberish at the turnover. That could give the ring setting for the fast rotor. I cannot find the ring setting for the middle rotor unless I have a turnover on that rotor. Same for the slow rotor. A few hundred characters should give me the middle rotor, and several thousand characters the slow rotor.
Sales suggests the different turnover positions would give away the rotor. If I know the message indicator is zzA, and I can somehow detect a turnover at zzR (Royal), then I know the rotor is I. But how do I detect the turnover?
The ring setting and the plugboard protect a message indicator sent in the clear. Without rings or plugboard, then one would just try all the rotor orders at the message indicator setting.
The plugboard is a monoalphabetic substitution. Is the M4 project hillclimbing just for calculating that with mono-, bi-, and tri-grams?
The German forms.
I think the Herivel tip article leads to a comment by Alexander about its value. The implication is Germans added complexity, but the complexity was incrementally doable. If there had been some major changes, then they would have defeated the attacks.
The Polish bomby relied on message indicator repetition. That system was working fine, but Turing thought it wouldn't last. The British bombe was designed to look for cribs, and it was already being designed/built (check) when the indicator repetition stopped.
Sales has the pinches as necessary because the tables could not be figured out.
The HT may be even more clever. Guess rotor order, but the tip gives both ring settings and rotor positions.
Company had nice pictures, but made some dubious claims.
Want frequency characteristics. GBW product.
Blue glow. Can gas breakdown law infer a level of vacuum in the audion? (May not be a glow discharge?)
Ultraaudion, audio converter, adjust filament power, "hissing", imperfect vacuum, shot noise oscillator "works" like Poulsen arc, "siren effect".
The Quarterly Journal of the University of North Dakota, Volume 5, 1915, http://books.google.com/books?pg=PA198&lpg=PA198&dq=frequency+characteristics+audion&sig=sokxdnjRiHXt0hXN6C0j4nosgIM&ei=Ewe6UfPvEcb_igKDoYCQAQ&id=kN8hAAAAMAAJ&ots=P1RqY-boCa&output=text
Several early radio publications have explanation. NBS explanation of three modes, only one important.
Nobody gives a good explanation of the IV characteristic. Unidirectional. Carbon electrode can get hot and have thermionic emission. Metal electrode is cooled, so poor thermionic emission. Magnetic blowout.
Ion issue. Looking for gas breakdown to arc but then ability to quench the arc?
IRE Vol 5. P. O. Pedersen, "On the Poulsen Arc and its Theory" date=August 1917 volume=5 issue=4 page 255
At Hathi Trust? http://catalog.hathitrust.org/Record/008616679
Vol 5 also has Armstrong, A Study of Heterodyne Amplification by the Electron Relay, pp 145
Yagi on transformers, page 433
The typical GDT has a low pressure noble gas. For the present, assume that the electrodes do not emit any electrons. Also assume there are no ions.
Consequently, the GDT is an insulator; no current flows.
Considering current density as the controlling variable, the voltage-current characteristic of a dc discharge is shown in Fig. 14.2. The flat region with slightly negative slope dV/dI is that of the normal glow. From low currents, the region below IA is called a dark or Townsend discharge. The glow gradually builds up until a transition is reached, with hysteresis, entering the normal glow at a voltage VS. The voltage remains constant as the current increases until IB, at which point there is an increasing voltage-current characteristic called the abnormal glow. A further increase in current resutls in a rather abrupt transistion at IC, again characterized by hysteresis, to a considerable lower voltage discharge known as an arc discharge.))
There will be some ions created by light, cosmic rays, or radio activity. Consequently, there will be some positive ions and electrons. With no applied voltage, there is no significant drift and they recombine. Applying a voltage causes the heavy positive ions to drift toward the cathode and the light electrons to drift to the anode. Some could still recombine.
Increasing the applied voltage causes all the generated ions to be swept to the anode and cathode before they recombine. Consequently the current reaches a saturation level of IS that stays constant as the voltage is increased.
See Rutherford (1899, pp. 112–113).
At some point, the drifting ions gather enough energy that they can generate more carriers. This region is Townsend discharge, and there are two dominant mechanisms. The first mechanism is electrons colliding with neutral atoms to generate an additional positive ion and electron. The positive ion drifts to the cathode, and the negative electron drifts to the anode. Avalanche multiplication is happening. Now there are two electrons, and each of them may gain enough energy to kick loose more electrons.
Discuss photons... Start of glow.
The second mechanism is secondary emission. When a positive ion crashes into the cathode, it may liberate another electron that will then drift to the anode (and, possibly, free more carriers).
Uncontrolled generation.
Switch modes.
Suggestion that secondary emission is now 1 for 1 at ignition / strike.
Extinction voltage.
GE reference says negative resistance, but Tube reference does not.
Local heating of the electrodes makes it easier to develop a current.
Voltage becomes similar to ionization potential.
Arc discharge requires electrode emission. Thermionic or high field.
Self destruction.
Negative resistance here.
Cathode dark space.
Striations.
Faraday dark space.
Column.
Anode glow.
Has lots of detail, but does not show reorganization of gas. Has negative resistance region at Townsend breakdown; acute angle.
Ott may have referred to glow region as Townsend discharge. Has an arc being a metal-arc discharge and suggesting that metal ions are involved.
I disagree with the notion that devices such as a gas discharge tube exhibit negative resistance.
Consider a non conducting gas tube just below the strike voltage. Now increase by one volt to fire the tube. If the terminal voltage is held constant, then the current increases dramatically instead of falling (as it would in a negative resistance).
If the terminal voltage is then decreased, then the current also decreases. Decrease the voltage enough, and the tube quenches. Where is the negative resistance? I see a state change and hysterisis due to that change, but I don't see a negative resistance.
In contrast, if the same terminal measurements are made varying the voltage across a tunnel diode, there's a region where the voltage is increased but the current decreases.
If the independent variable is the current, then yes, there's a point where the current increases but the terminal voltage falls because the tube finally strikes. That transition, however, is also a discontinuous jump on the device IV curve. With the jump, there's no clear notion of a small signal impedance.
I don't see a nonlinearity or hysterisis implying a negative resistance.
Where's a reference that states a gas discharge tube is a negative resistance?
Does a circuit that can be broken somewhere in the feedback loop and display a negative resistance qualify? Negative resistance should not be just another word for power gain. That is the sense used in microwave oscillators, but the wizard behind the curtain is an amplifier.
Philbrick ref (page 8) says it resembles a Howland with infinite load.
In addition to Ott...
History of metal arcs. 1903; rediscovered. Voltage less than 6 volts to avoid breakdown.
For good background, see TranslateWiki.net FAQ.
Conditionals http://www.mediawiki.org/wiki/Help:Extension:ParserFunctions
Rounding
Grammar doesn't do anything on en.WP? Try on de.WP for declension.
Related plural magic word:
Documentation is wrong. See table at end of documentation.
((cite))
using tl2((cite))
using tl2 with sister=simple((cite))
using tlx with sister=simple (sister does not work)((cite))
using tltt with sister=simple (sister does not work)See ((Coord)) at Helper functions. There's also Wikipedia:WikiProject Geographical coordinates#Creating new templates
The current language is [undefined] Error: ((Lang)): no text (help). (another Commons issue?)
Statement in another language: Ich bin ein Berliner. (I am a jelly donut.)
This page contains a translation of German sentence. from de.wikipedia. |
LangSwitch (Commons?) is not connecting ((LangSwitch|en=English trouble |de=German trouble |fr=French trouble |default=Default trouble)):English trouble
Language
Pretty basic, but has other citations:
((citation))
: Cite has empty unknown parameter: |inventor1-first=
(help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor1-last=
ignored (help); Unknown parameter |patent-number=
ignored (help)Autoload but manually cocked.
((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor1-first=
ignored (help); Unknown parameter |inventor1-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Drawing is "Magazine Firearm".
((citation))
: Unknown parameter |assign1=
ignored (help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Title on drawing is "Magazine Fire Arm".
((citation))
: Check date values in: |publication-date=
(help); Unknown parameter |assign1=
ignored (help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Title on drawing is "Magazine Fire-Arm".
I'm confused about the models. The simple explanation appears to be 1866 (rimfire) is without the safety, the 1873 has a safety, and the 1876 has a longer receiver for the longer cartridges. Winchester could not make the 1876 strong enough for the government cartridge, so it used a slightly different cartridge.
Henry rifle; rimfire .44 Henry rifle cartridge. Had a follower but no forestock. Manufactured 1860 to 1866. Apparently iron or brass receiver. Iron first, brass second.
Winchester 1866. Still rimfire? King gate. Yellow boy.
Winchester 73. Pictures vary all over the map. Don't believe it had a safety. Iron and then steel action.
Winchester 76. With safety. longer receiver for longer cartridges.
But page 228A shows an 1876 in .44/40 (something that Sharpe does not list as a cartridge for the '76.) The receiver is longer and it has a safety.
((citation))
: Check date values in: |publication-date=
(help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help) (guess from image on net)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor1-first=
ignored (help); Unknown parameter |inventor1-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor1-first=
ignored (help); Unknown parameter |inventor1-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help), gas action for machine guns((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor1-first=
ignored (help); Unknown parameter |inventor1-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help), breech mechanism of machine gunDescribe the gain mechanism.
Grit versus knurled wheel printer controversy. "B & L agrees that the use of grit provides great advantages over a knurled wheel. In particular, grit is much harder and sharper than the points of a knurled wheel, and so is especially effective in creating indentations in the paper." (909 F.2d 1464 @ ¶ 16.)
LaBarre was an "apparatus" patent. Differences in performance are not relevant. (¶ 18):
Interesting machine that winds toroid cores. Basic plan is to get a loop of wire rotating through toroid; coil falls off loop and gets wrapped around the toroid; toroid spins to evenly distribute the wire.
Baker (1959) uses a soldering iron (for starting?). I don't see slip off is regulated. Loop is distored to travel through toroid.
Oshima (1964) has some slip cups above and below the toroid (10, 10'). Slip may be through bobbin.
Potthoff (1978) Hand crank, drive belt pulled inside so loop can travel through torid.
((citation))
: Unknown parameter |air-date=
ignored (help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |air-date=
ignored (help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |air-date=
ignored (help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |air-date=
ignored (help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Hand draw figures. Improve directivity. Tektronix.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Rhode-Schwarz.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Stanford University. 100GHz.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Hewlett-Packard. Earlier of 4720677.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Hewlett-Packard((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). VNA. Hewlett-Packard. Also use FFT to get time domain. 8510?((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-first=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Spectrum analyzer. Older VNA? Hewlett-Packard. Was cited in US 4816767.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). doubled bridge? GTE Automatic Electric Lab.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Test set. 300kHz to 3.0GHz. Hewlett-Packard.It's a small world. I was looking at spot welding, and someone provided a early Popular Mechanics article that made me wonder who invented spot welding. A long time later, I'm looking at oscillators and come across Elihu Thomson with a singing arc -- something I'd heard about in my youth. Then I find out he invented resistance welding:
((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventorlink=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventorlink=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help)
((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). A capacitor (aka condenser) is used to limit the energy delivered during a spot welding operation. Capacitor is charged rather than discharged. Goal is delicate work.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Second condenser is discharged into the weld. "I have discovered that the weld is considerably improved if a second condenser is provided and connected in such a manner that it discharges into the weld...."((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Capacitive discharge with mechanical energy storage and step down transformer. Goal is high peak powers.((citation))
: Cite has empty unknown parameter: |inventor-first=
(help); Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Construction details for welding transformer.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Capacitive discharge spot welder with stepdown transformer. Goal is delicate work.((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help).((citation))
: Unknown parameter |country-code=
ignored (help); Unknown parameter |inventor-first=
ignored (help); Unknown parameter |inventor-last=
ignored (help); Unknown parameter |inventor2-last=
ignored (help); Unknown parameter |issue-date=
ignored (help); Unknown parameter |patent-number=
ignored (help). Goal is welding seams that have a resistance in the 1 to 100 micro-ohm range.For just pages=, a hyphen is changed to an endash. U+2011 or #8209; is a nonbreaking hyphen. (- ‑ − – —)
Scanned halftone images have problems. Moiré patterns.
There was a paper by Picture Elements, Inc. about filterning the images, but I cannot find it. DFTs found the screen frequency and screen angle.
Another PEI document
Advice on web has stuff about Gaussian blur or other image processing filters in programs such as Photoshop.
((cite journal))
: Cite journal requires |journal=
(help)It is desirable to have all XML files validate. It was even a goal of writing DTDs. Some DOM parsers will raise an exception if the DTD is violated. (They also take more compute time to do the validation. I had to find out how to turn off Java's XML validation because it was taking 6 seconds to validate a file.) Even if validation is not done all the time, it is reassuring to know that an XML file validates.
SVG files should display ((Valid SVG)), but that is where some trouble arises. Unfortunately, including information (such as Common Core licensing) in a metadata
element often prevents validation because the DTD is strict. Furthermore, XML extensions are often sprinkled throughout an XML file. Translation markup often want to identify elements that should not be translated with the attribute its:translate="no"
. That attribute is not in the underyling DTD, so it raises an error. One way around the problem is to extend the underlying DTD; HTML 5, for example, now has a translate
attribute. That's not universal; SVG 2.0 does not have a translate
attribute. Many XML-based file formats now recommend against using a DOCTYPE
declaration. Apparently, a DTD was never written for SVG 1.2. DTD were also developed before XML namespaces. Namespaces give name isolation, but they confound the DTD issue.
Look at schemas. Relax.
IIRC, there are some namespace twists. Elements without a namespace prefix belong to the default namespace (whatever xmlns=...
dictates), but attributes without a namespace prefix belong to the null namespace.
Many experiments provide insight.
Many XML validators are available. DOM processors have internal validators. There are also web-based validators. W3 has one at
((Valid SVG)) uses the W3 validator, so it deserves some attention.
which provides the following example (validates as XML due to SYSTEM; added encoding="UTF-8"):
<?xml version="1.0" encoding="UTF-8" ?>
<!DOCTYPE rdf:RDF SYSTEM "http://dublincore.org/2000/12/01-dcmes-xml-dtd.dtd">
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:dc ="http://purl.org/dc/elements/1.1/">
<rdf:Description rdf:about="http://dublincore.org/">
<dc:title>Dublin Core Metadata Initiative - Home Page</dc:title>
<dc:description>The Dublin Core Metadata Initiative Web site.</dc:description>
<dc:date>1998-10-10</dc:date>
<dc:format>text/html</dc:format>
<dc:language>en</dc:language>
<dc:contributor>The Dublin Core Metadata Initiative</dc:contributor>
<dc:contributor xml:lang="fr">L'Initiative de métadonnées du Dublin Core</dc:contributor>
<dc:contributor xml:lang="de">der Dublin-Core Metadata-Diskussionen</dc:contributor>
</rdf:Description>
</rdf:RDF>
Metadata version that fails SVG 1.1 validation (validates as XML if DOCTYPE removed; does not complain about <gubbish/>
tag, so no SVG validation done!):
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd" >
<svg version="1.1" baseProfile="full"
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:ev="http://www.w3.org/2001/xml-events"
xmlns:its="http://www.w3.org/2005/11/its"
width="1200"
height="500"
viewBox="0 0 1200 500"
overflow="visible"
enable-background="new 0 0 1200 500"
xml:space="preserve">
<title>Energy to ionize the first electron</title>
<desc>Energy required to ionize the first electron from an atom.</desc>
<metadata id="license"
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:cc="http://creativecommons.org/ns#">
<rdf:RDF>
<cc:Work rdf:about="http://commons.wikimedia.org/wiki/File:First_Ionization_Energy.svg">
<dc:format>image/svg+xml</dc:format>
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
<cc:license rdf:resource="http://creativecommons.org/licenses/by-sa/3.0/" />
</cc:Work>
<cc:License rdf:about="http://creativecommons.org/licenses/by-sa/3.0/">
<cc:permits rdf:resource="http://creativecommons.org/ns#Reproduction" />
<cc:permits rdf:resource="http://creativecommons.org/ns#Distribution" />
<cc:requires rdf:resource="http://creativecommons.org/ns#Notice" />
<cc:requires rdf:resource="http://creativecommons.org/ns#Attribution" />
<cc:permits rdf:resource="http://creativecommons.org/ns#DerivativeWorks" />
<cc:requires rdf:resource="http://creativecommons.org/ns#ShareAlike" />
</cc:License>
</rdf:RDF>
</metadata>
</svg>
Another validator
Sometimes the validator decides the SVG is SVG 1.1+XHTML+MathML 3.0 and ignores Inkscape and RDF, but it didn't for me in the sections below. I copied the header and tried the stuff immediately below, but it was just checked for XML.
I took a file that validates as SVG 1.1+XHTML+MathML 3.0, copied its contents into the clipboard, and then did a direct input validation. It validated as XML.
I suspect there is no DTD for SVG 1.1+XHTML+MathML 3.0. Consequently, there is no DOCTYPE that can be entered to specify that checking. Instead, the validator probably cues off the webserver's Content-Type
. With direct input, the validator may assume text/xml
rather than something like image/svg+xml
. Checked that WP returns image/svg+xml
, so I believe this hypothesis.
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!-- Created with Inkscape (http://www.inkscape.org/) -->
<svg xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:cc="http://creativecommons.org/ns#"
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:svg="http://www.w3.org/2000/svg"
xmlns="http://www.w3.org/2000/svg"
xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
version="1.1" width="958.69" height="592.78998" id="svg2275"
inkscape:version="0.48.2 r9819"
sodipodi:docname="Map_of_USA_with_state_names-en.svg">
<metadata id="metadata362">
<rdf:RDF>
<cc:Work rdf:about="">
<dc:format>image/svg+xml</dc:format>
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
</cc:Work>
</rdf:RDF>
</metadata>
</svg>
The code from SVG 1.1 spec (http://www.w3.org/TR/SVG/metadata.html) validates as XML (no SVG DOCTYPE).
<?xml version="1.0" standalone="yes"?>
<svg width="4in" height="3in" version="1.1"
xmlns = 'http://www.w3.org/2000/svg'>
<desc xmlns:myfoo="http://example.org/myfoo">
<myfoo:title>This is a financial report</myfoo:title>
<myfoo:descr>The global description uses markup from the
<myfoo:emph>myfoo</myfoo:emph> namespace.</myfoo:descr>
<myfoo:scene><myfoo:what>widget $growth</myfoo:what>
<myfoo:contains>$three $graph-bar</myfoo:contains>
<myfoo:when>1998 $through 2000</myfoo:when> </myfoo:scene>
</desc>
<metadata>
<rdf:RDF
xmlns:rdf = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:rdfs = "http://www.w3.org/2000/01/rdf-schema#"
xmlns:dc = "http://purl.org/dc/elements/1.1/" >
<rdf:Description rdf:about="http://example.org/myfoo"
dc:title="MyFoo Financial Report"
dc:description="$three $bar $thousands $dollars $from 1998 $through 2000"
dc:publisher="Example Organization"
dc:date="2000-04-11"
dc:format="image/svg+xml"
dc:language="en" >
<dc:creator>
<rdf:Bag>
<rdf:li>Irving Bird</rdf:li>
<rdf:li>Mary Lambert</rdf:li>
</rdf:Bag>
</dc:creator>
</rdf:Description>
</rdf:RDF>
</metadata>
</svg>
Play some games...
Probably should not say standalone="yes"
. Problems with defaulted from external subset.
Gets one error -- does this mean metadata element is not ANY? The metadata
rule is (#PCDATA ...)
Line 25, Column 58: document type does not allow element "rdf:RDF" here xmlns:dc = "http://purl.org/dc/elements/1.1/" >✉ The element named above was found in a context where it is not allowed. This could mean that you have incorrectly nested elements -- such as a "style" element in the "body" section instead of inside "head" -- or two elements that overlap (which is not allowed). One common cause for this error is the use of XHTML syntax in HTML documents. Due to HTML's rules of implicitly closed elements, this error can create cascading effects. For instance, using XHTML's "self-closing" tags for "meta" and "link" in the "head" section of a HTML document may cause the parser to infer the end of the "head" section and the beginning of the "body" section (where "link" and "meta" are not allowed; hence the reported error).
<?xml version="1.0" ?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"
[
<!ELEMENT rdf:RDF ANY>
<!ELEMENT rdf:Description ANY>
<!ATTLIST rdf:Description dc:title CDATA #IMPLIED
dc:description CDATA #IMPLIED
dc:publisher CDATA #IMPLIED
about CDATA #IMPLIED
dc:format CDATA #IMPLIED
dc:language CDATA #IMPLIED
dc:date CDATA #IMPLIED>
<!ELEMENT rdf:Bag ANY>
<!ELEMENT rdf:li ANY>
<!ELEMENT dc:creator ANY>
<!-- and the trick... first definition wins -->
<!ENTITY % SVG.metadata.extra.content "| rdf:RDF" >
]>
<svg width="4in" height="3in" version="1.1"
xmlns = 'http://www.w3.org/2000/svg'>
<desc>Just some simple text</desc>
<metadata>
<rdf:RDF
xmlns:rdf = "http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:rdfs = "http://www.w3.org/2000/01/rdf-schema#"
xmlns:dc = "http://purl.org/dc/elements/1.1/" >
<rdf:Description rdf:about="http://example.org/myfoo"
dc:title="MyFoo Financial Report"
dc:description="$three $bar $thousands $dollars $from 1998 $through 2000"
dc:publisher="Example Organization"
dc:date="2000-04-11"
dc:format="image/svg+xml"
dc:language="en" >
<dc:creator>
<rdf:Bag>
<rdf:li>Irving Bird</rdf:li>
<rdf:li>Mary Lambert</rdf:li>
</rdf:Bag>
</dc:creator>
</rdf:Description>
</rdf:RDF>
</metadata>
</svg>
For SVG data-*
attributes, extend the SVG 1.1 DTD attributes with "SVG.External.extra.attrib".
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"
[
<!ENTITY % SVG.External.extra.attrib "
data-area CDATA #IMPLIED
data-src CDATA #IMPLIED
">
]>
There can be problems within a file that make it fail validation. For example, problems include invalid NCNAMEs in identifier attributes. An interesting problem was unnormalized Unicode strings raising errors for an SVG file. There are methods than can used to normalize Unicode strings.
See extensive discussion on Commons:User_talk:Delphi234.
That file passes SVG validation after several fixes.
switch
elements (i.e., bare characters rather than within text
elements. (deleted extra characters)title
element<g xml:space="preserve" style="font-size:13.08702374px;font-style:normal;font-variant:normal;font-weight:bold;font-stretch:normal;text-align:start;line-height:100%;writing-mode:lr-tb;text-anchor:middle;fill:#000000;fill-opacity:0.75;stroke-width:3pt;font-family: Nimbus Sans L,'Arial Narrow',sans-serif;Sans L',sans-serif;-inkscape-font-specification:Nimbus Sans L Bold">
Font shifts needed? bn versus bpy
There is a large issue about licensing. I do not understand why wiki files do not embed the license terms and a link to Commons in its exported files that support such information. TIFF, JPEG/Exif, and SVG/XML can carry metadata and linking.
Sadly, I've seen somebody take a Commons file, edit it, and upload it with a claim that it is their own. The metadata should be scavenged from the source file (and perhaps not be alterable). If there is no metadata, then it should be inserted. If a file is copied from Commons, it should be able to phone home.
CC watermarks appear discouraged.
Some activity on commons: Commons:Commons:Structured data. de.WP has funding....
There's a category issue. Maybe get categories from original and then filter?
Ah, there is a difference between uploading a new file and updating an existing file.
On meta:
WMF wants free images to stay free, but that seems in conflict with typical practice.
What was the name of the Commons proposal? There was a recent RfC with de.WP having source of funding.
See
Let Qxxxx be a copyrightable work.
Many illustrations have been credited to a book's author rather than the illustrator.
In addition, many illustrations are missing ((PD-scan))
.
Walther Otto Müller (Walther Otto Müller (Q21521878))
Otto Wilhelm Thomé Otto Wilhelm Thomé (Q76714)
No Q for Franz Köhler the botanist.
The Musée d'Orsay:
It describes the painting
So I get the creation date and the date of death. 120 years after death is 1997. France sweat-of-the-brow copyright cases have gone both ways.
Metadata is a broad topic, but the basic notion is putting information, such as exposure, location, and copyright information, into the file that contains the primary data. Most picture formats have tags for metadata.
The primary focus is putting copyright information into an rdf:RDF
element inside of SVG's metadata
element in such a way that the SVG file will still validate.
A secondary focus would be adding ITS information, but that will probably cause validation to fail.
Dublin core defines the vocabulary for representing relationships such as author.
Primary license expression on Wikipedia follows Creative Commons and is usually expressed as a Creative Commons license. The license variants are expressed by some (possibly exclusive) properties:
There are different versions of these licenses (including ported versions which are disappearing). URIs exist for different licenses:
These URIs link to HTML descriptions of the licenses. I expect to get an RDF description, but instead there is a link that I'd expect to be to a straight RDF description. However, I cannot find the linked file.
<link rel="alternate" type="application/rdf+xml" href="rdf">
How these licenses are claimed in different files is covered in ccREL: The Creative Commons Rights Expression Language :
It covers several formats such as HTML, RDF, and SVG.
<div about="http://lessig.org/blog/" xmlns:cc="http://creativecommons.org/ns#">
This page, by
<a property="cc:attributionName" rel="cc:attributionURL" href="http://lessig.org/">Lawrence Lessig</a>,
is licensed under a
<a rel="license" href="http://creativecommons.org/licenses/by/3.0/">Creative Commons Attribution License</a>.
</div>
ccREL describes the rights as a work and a license.
Separate RDF files.
The representation is now Work + possibly a license.
Work triples/properties
License triples/properties
Specify value with rdf:resource
attribute or element contents.
<metadata>
<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:cc="http://creativecommons.org/ns#">
<cc:Work rdf:about="">
<dc:title>Diagram Verfassung Bosch</dc:title>
<dc:source rdf:resource="https://commons.wikimedia.org/wiki/File:Diagram_Verfassung_Bosch.png"/>
<cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Habakuk"/>
<cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Lasse_Havelund"/>
<cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Perhelion"/>
<cc:attributionName rdf:resource="https://commons.wikimedia.org/wiki/User:Glrx"/>
<cc:attributionURL rdf:resource="https://commons.wikimedia.org/w/index.php?title=File%3ABosch_Composition.svg"/>
<dc:format>image/svg+xml</dc:format>
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
<cc:license rdf:resource="http://creativecommons.org/licenses/by-sa/3.0/deed.en" />
</cc:Work>
</rdf:RDF>
</metadata>
RDFa.
<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd" >
<svg version="1.1" baseProfile="full"
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:ev="http://www.w3.org/2001/xml-events"
xmlns:its="http://www.w3.org/2005/11/its"
width="1200"
height="500"
viewBox="0 0 1200 500"
overflow="visible"
enable-background="new 0 0 1200 500"
xml:space="preserve">
<title>Energy to ionize the first electron</title>
<desc>Energy required to ionize the first electron from an atom.</desc>
<metadata id="license"
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:cc="http://creativecommons.org/ns#">
<rdf:RDF>
<cc:Work rdf:about="http://commons.wikimedia.org/wiki/File:First_Ionization_Energy.svg">
<dc:format>image/svg+xml</dc:format>
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
<cc:license rdf:resource="http://creativecommons.org/licenses/by-sa/3.0/" />
</cc:Work>
<cc:License rdf:about="http://creativecommons.org/licenses/by-sa/3.0/">
<cc:permits rdf:resource="http://creativecommons.org/ns#Reproduction" />
<cc:permits rdf:resource="http://creativecommons.org/ns#Distribution" />
<cc:requires rdf:resource="http://creativecommons.org/ns#Notice" />
<cc:requires rdf:resource="http://creativecommons.org/ns#Attribution" />
<cc:permits rdf:resource="http://creativecommons.org/ns#DerivativeWorks" />
<cc:requires rdf:resource="http://creativecommons.org/ns#ShareAlike" />
</cc:License>
</rdf:RDF>
</metadata>
</svg>
Embedded RDF. There's a standard for XSL transformation to extract the embedded RDF.
<?xml version="1.0" encoding="UTF-8" standalone="no"?>
<!-- Created with Inkscape (http://www.inkscape.org/) -->
<svg xmlns:dc="http://purl.org/dc/elements/1.1/"
xmlns:cc="http://creativecommons.org/ns#"
xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
xmlns:svg="http://www.w3.org/2000/svg"
xmlns="http://www.w3.org/2000/svg"
xmlns:sodipodi="http://sodipodi.sourceforge.net/DTD/sodipodi-0.dtd"
xmlns:inkscape="http://www.inkscape.org/namespaces/inkscape"
version="1.1" width="958.69" height="592.78998" id="svg2275"
inkscape:version="0.48.2 r9819"
sodipodi:docname="Map_of_USA_with_state_names-en.svg">
<metadata id="metadata362">
<rdf:RDF>
<cc:Work rdf:about="">
<dc:format>image/svg+xml</dc:format>
<dc:type rdf:resource="http://purl.org/dc/dcmitype/StillImage"/>
</cc:Work>
</rdf:RDF>
</metadata>
</svg>
There's got to be an Resource Description Framework (RDF/XML) editor around somewhere.
There's a file on Commons, and somebody copies it and uses it. There's no guarantee of attribution. There could be RDF info, and some of that could be automatically inserted. Something basic would just make a URL point back to Commons. Further efforts would make sure the file and the Commons description refer to the same sharing license.
Consider the tag set.
Also, JPEGs should be checked their Exchangeable image file format copyright information, but I'm confused about this. Copyright appears to be in TIFF and WAV files, but I'm not clear about JPEG. It may be that TIFF style information is placed in the JPEG#Syntax and structure file as APP1
.
Several comments on the web about no RDF DTD. Here's an old DTD attempt:
which, if examined, says the following:
<!--
This DTD fragment is identified by and should be available at
http://www.w3.org/Metadata/RDF/Group/9710/rdf.dtd
Copyright 1997, 1999 by MIT, INRIA, and Keio
$Id: rdf.dtd,v 1.3 1999/02/08 17:41:51 connolly Exp $
It is a transcription of the grammar in
Resource Description Framework (RDF) Model and Syntax
Ora Lassila and Ralph R. Swick
Version 0.2, 01-Oct-1997
http://www.w3.org/Metadata/RDF/Group/9710/WD-rdf-syntax-971002
updated for
http://www.w3.org/TR/1999/PR-rdf-syntax-19990105
The grammar is presented in a different order because SGML
requires parameter entities to be declared before they are referenced.
Typical usage:
<!DOCTYPE rdf:RDF system
"http://www.w3.org/XML/9710rdf-dtd/rdf.dtd">
<rdf:RDF>...</rdf:RDF>
Documents that include this DTD fragment by reference
SHOULD NOT reference parameter entities declared herein
(e.g. %RDF;).
These names are intended to be "private."
Only the element and attribute names are intended to be exported.
The parameter entity rdf-namespace-prefix is intended to
be "imported" ala:
<!DOCTYPE [
<!ENTITY % rdf-namespace-prefix "blort:">
<!ENTITY % rdf-dtd system
"http://www.w3.org/XML/9710rdf-dtd/rdf.dtd">
%rdf-dtd;
]>
<blort:RDF>...</blort:RDF>
Oh for real SGML/XML namespaces... ;-)
-->
References
GRDDL ("Griddle") is used to extract RDF from XML files. Gleaning Resource Descriptions from Dialects of Languages.
RDF can be used in maps to identify coordinate transformations. RDF can also be used to designate a past period.
Take, for example, an old map of Port Royal.
Documents
The Wikimaps warper is not using RDF. After warping, the original image on Commons is unchanged. There is a link to the warper page that has a mapping table.
Want a query that will grab nearby cities. Filter on longitude and latitude? long 8 to 28; lat 41 to 52
SELECT DISTINCT ?city ?cityLabel (SAMPLE(?location) AS ?location) (MAX(?population) AS ?population)
WHERE
{
?city wdt:P31/wdt:P279* wd:Q515;
wdt:P625 ?location;
wdt:P1082 ?population.
FILTER(?population >= 500000).
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
GROUP BY ?city ?cityLabel
SELECT ?place ?placeLabel ?location
WHERE
{
# Budapest coordinates
wd:Q1781 wdt:P625 ?Loc .
SERVICE wikibase:around {
?place wdt:P625 ?location .
bd:serviceParam wikibase:center ?Loc .
bd:serviceParam wikibase:radius "300" .
} .
# Is a city - now human settlement
FILTER EXISTS { ?place wdt:P31/wdt:P279* wd:wd:Q486972 } .
?place wdt:P1082 ?population .
FILTER (?population > 50000) .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en" .
}
}
I thought that Unicode would simplify the translation problem: one could use a single font for all the translations. Turns out, different languages/locales prefer different character shapes. That means that some translations should change fonts or do character selection within a font.
See CSS Fonts Module Level 4.0. Uses OpenType language names: https://docs.microsoft.com/en-us/typography/opentype/spec/languagetags
Allegedly Firefox only, but cannot get it to work.
Usually, the content language specification is enough.
Sometimes font-language-override
is used to render an unsupported content language with a similar language.
Same characters (I do not see a difference)
АаБбВвГгДдЕеЁёЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъЫыЬьЭэЮюЯя
АаБбВвГгДдЕе ЖжЗзИиЙйКкЛлМмНнОоПпРрСсТтУуФфХхЦцЧчШшЩщЪъ Ьь ЮюЯя
Same word in Spanish Señora and Italian Señora.
Try historical forms (looking for Jeſuit and Congreſs):
See Long s.
Unicode: LATIN SMALL LETTER LONG S ſ: "ſ", so Jeſuit and Congreſs.
Such a change could be done with CSS, but that gets us into different problems.
The CSS :lang
selector (e.g., p:lang(zh) {..}
).
The xml:lang
and HTML lang
attributes should track.
Should make some test code to see if CSS looks at both attributes.
HTML lang
attributes are single-valued.
HTML 5 seems to want xml:lang
to disappear in favor of lang
.
SVG is more complicated (and CSS rule may not be available).
The xml:lang
attribute is single valued.
The SVG lang
attribute appears in SVG 2.0 and is single valued.
The systemLanguage
attribute is multi-valued — a reason to shatter the attribute.
SVG switch
uses the systemLanguage
attribute.
First consider single-valued attribute solutions.
That could use attribute value selection to set the font.
Alternatively, CSS to install xml:lang
attribute depending on systemLanguage
attribute (check).
Alternatively, one could always set a language attribute on nodes that need font selection.
(There should be an ITS rule that avoids translating children of a switch
.)
XLIFF localization.
Illustrations can avoid the segmentation step because SVG strings will usually be short, but title and desc elements may be exceptions.
There's a master file (say in .en). Then it is processed via .en.de XLIFF. Also a .en.es XLIFF.
There can be IDs, but they should be unique to the file. The lookup by ID. I want IDs that identify the content rather than the instance.
XLIFF filter: Take multilanguage SVG and convert to XLIFF.
The trivial version is just an XSLT that grabs the text
nodes.
Slightly better version would pay attention to xml:lang
, its:translate
, and ITS rules.
ITS rules could have "switch/text" and "@systemLanguage" awareness. The "@systemLanguage" attribute is further complicated because it is a list rather than just one IETF language code; see if XSLT can match substring patterns.
Need a second filter that produces the skeleton, and a third filter to merge XLIFF to language version.
Interaction of XLIFF and other schemes with Internationalization Tag Set.
xml:lang
attribute. HTML has compatible lang
attribute that has consistency rules (if both exist, they must be the same).its:translate
attribute. HTML 5 has compatible global attribute translate
.its:locNote
, its:locNotePointer
, its:locNoteRef
, its:locNoteRefPointer
its:term
attribute. HTML 5 does not have this attribute. (http://www.w3schools.com/tags/ref_attributes.asp)
its:termRule
requires one of its:termInfoRef
(URI), its:termInfoPointer
(XPath), or its:termInfoRefPointer
(XPath)its:termConfidence
2.0Question: does SVG 2.0 have translate
attribute? Apparently not.
There are data-*
attributes in HTML 5 and SVG 2.
There are its-...
attributes in HTML 5 (or at least the Nu validator).
It makes local notes trivial. HTML with those attributes (camel-case conversion hack) will validate in Nu.
<!DOCTYPE html>
<html lang=en>
<head>
<meta charset=utf-8>
<title>Terminology test: default</title>
</head>
<body>
<p>We need a new <span its-term=yes>motherboard</span>
</p>
</body>
</html>
I don't get the same luck when the its
namespace is introduced.
The Nu validator complains about introducing the new namespace and the new attribute.
Of course, the validator does not have a specification for ITS.
<!DOCTYPE html>
<html lang="en" xmlns:its="http://www.w3.org/2005/11/its" its:version="2.0" >
<head>
<meta charset=utf-8>
<title>Terminology test: default</title>
</head>
<body>
<p>We need a new <span its-term=yes>motherboard</span>
</p>
</body>
</html>
Now try with SVG. Augment the SVG 1.1 DTD to include some ITS definition.
Rules are stuffed in metadata
element.
Local markup stuffed everywhere, but it could be tighter.
<?xml version="1.0" encoding="utf-8" standalone="no" ?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd"
[
<!-- Miniature DTD for ITS -->
<!ELEMENT its:rules (its:termRule
| its:translateRule
| its:withinTextRule
| its:locNoteRule
| its:domainRule
)* >
<!ATTLIST its:rules version CDATA #REQUIRED
queryLanguage CDATA #FIXED "xpath"
xlink:href CDATA #IMPLIED
xlink:type (simple) #IMPLIED >
<!ELEMENT its:termRule EMPTY >
<!ATTLIST its:termRule selector CDATA #REQUIRED >
<!ELEMENT its:translateRule EMPTY >
<!ATTLIST its:translateRule selector CDATA #REQUIRED
translate (yes | no) #REQUIRED >
<!ELEMENT its:withinTextRule EMPTY >
<!ATTLIST its:withinTextRule selector CDATA #REQUIRED
withinText (yes | no) #REQUIRED >
<!ELEMENT its:locNoteRule (its:locNote)? >
<!ATTLIST its:locNoteRule selector CDATA #REQUIRED
locNoteType (description | alert) #REQUIRED
locNotePointer CDATA #IMPLIED
locNoteRef CDATA #IMPLIED
locNoteRefPointer CDATA #IMPLIED >
<!ELEMENT its:locNote (#PCDATA) >
<!ELEMENT its:domainRule EMPTY >
<!ATTLIST its:domainRule selector CDATA #REQUIRED
domainPointer CDATA #REQUIRED
domainMapping CDATA #IMPLIED >
<!-- Include those elements in the SVG external subset -->
<!ENTITY % SVG.metadata.extra.content "| its:rules" >
<!ENTITY % SVG.External.extra.attrib "
its:version CDATA #IMPLIED
its:term (yes | no) #IMPLIED
its:termInfoRef CDATA #IMPLIED
its:termConfidence CDATA #IMPLIED
its:translate (yes | no) #IMPLIED
its:locNote CDATA #IMPLIED
its:locNoteRef CDATA #IMPLIED
its:locNoteType (description | alert) 'description'
its:withinText (yes | no | nested) 'no'
">
]>
<svg version="1.1"
viewBox="0 0 640 320"
xml:lang="en"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:its="http://www.w3.org/2005/11/its"
its:version="2.0" >
<title>Terminology test: default</title>
<metadata>
<its:rules version="2.0">
<its:translateRule translate="no" selector="//term" />
<its:withinTextRule withinText="yes" selector="//tspan" />
</its:rules>
</metadata>
<text x="10" y="10" its:translate="yes">We need a new <tspan its:term="yes">motherboard</tspan>.</text>
</svg>
For SVG 1.2, there may be some ITS accommodations.
ITS 1.0 local markup
Quotation from the ITS v2.0 specification:
Local Data Categories. There are some distinctions in ITS v2.0 when used with HTML. HTML has local markup for the Translate, Directionality, and Language Information data categories.
Translation Memory eXchange is another XML format for translated pairs. Google Translate apparently accepts it. GT was free, but apparently was abused and is now a paid service.
{"responseData":{"translatedText":"Blockstripper","match":1},"responseDetails":"","responseStatus":200,"responderId":"235",
"matches":[
{"id":"441919990","segment":"stripper","translation":"Blockstripper","quality":"55","reference":" |@| ","usage-count":3,"subject":"Mechanical","created-by":"IATE","last-updated-by":"IATE","create-date":"2014-11-13 21:25:33","last-update-date":"2014-11-13 21:25:33","tm_properties":"[]","match":1},
{"id":"449281594","segment":"Stripper.","translation":"Vorrichtung zum Entfernen des Stanzabfalls.","quality":"55","reference":"http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/|@|http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/","usage-count":1,"subject":"Legal_and_Notarial","created-by":"Uni-heidelberg.de","last-updated-by":"Uni-heidelberg.de","create-date":"2014-11-28 17:25:00","last-update-date":"2014-11-28 17:25:00","tm_properties":"[]","match":0.98},
{"id":"451341627","segment":"Stripper","translation":"Abstreifeinrichtung","quality":"55","reference":"http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/|@|http:\/\/www.cl.uni-heidelberg.de\/statnlpgroup\/pattr\/","usage-count":1,"subject":"Legal_and_Notarial","created-by":"Uni-heidelberg.de","last-updated-by":"Uni-heidelberg.de","create-date":"2014-11-28 17:59:35","last-update-date":"2014-11-28 17:59:35","tm_properties":"[]","match":0.97}]}
<?xml version="1.0" ?>
<!DOCTYPE tmx SYSTEM "tmx11.dtd">
<tmx version="1.1">
<header
creationtool="MyMemory - mymemory.translated.net"
creationtoolversion="Beta"
segtype="sentence"
o-tmf="TW4Win 2.0 Format"
adminlang="EN-US"
srclang="EN-GB"
datatype="rtf"
creationdate="20160929T211941Z"
creationid="Translated">
</header>
<body>
<tu tuid="0" creationdate="20160929T211941Z" creationid="IATE" usagecount="3" lastusagedate="20160929T211941Z">
<tuv lang="EN-GB" creationdate="20160929T211941Z" creationid="IATE">
<seg>stripper</seg></tuv>
<tuv lang="DE-DE" creationdate="20160929T211941Z" creationid="IATE" changedate="20141113T212533Z" changeid="IATE">
<seg>Blockstripper</seg></tuv>
</tu>
<tu tuid="1" creationdate="20160929T211941Z" creationid="MyMemoryLoader" usagecount="2" lastusagedate="20160929T211941Z">
<tuv lang="EN-GB" creationdate="20160929T211941Z" creationid="MyMemoryLoader">
<seg>Stripper.</seg></tuv>
<tuv lang="DE-DE" creationdate="20160929T211941Z" creationid="MyMemoryLoader" changedate="20141128T172500Z" changeid="MyMemoryLoader">
<seg>Vorrichtung zum Entfernen des Stanzabfalls.</seg></tuv>
</tu>
<tu tuid="2" creationdate="20160929T211941Z" creationid="MyMemoryLoader" usagecount="2" lastusagedate="20160929T211941Z">
<tuv lang="EN-GB" creationdate="20160929T211941Z" creationid="MyMemoryLoader">
<seg>Stripper</seg></tuv>
<tuv lang="DE-DE" creationdate="20160929T211941Z" creationid="MyMemoryLoader" changedate="20141128T175935Z" changeid="MyMemoryLoader">
<seg>Abstreifeinrichtung</seg></tuv>
</tu>
</body>
</tmx>
Also Segmentation Rules eXchange.
Consider doing XML, SVG, and XLIFF stunts with multiple files and with Wiktionary.
Editor provides single language PNG file. Bitmaps are not scalable. They are difficult to manipulate. For other wikis, translation is needed. Translations require erasure and addition.
Editor provides single language SVG file. Bare minimum: quick and dirty word substitution. Don't have sentences.
Where is the link for the tool that edits SVG files to make new language versions?
Some to do
Consider the Gibraltar SVG file (a featured picture). Large SVG file, now many versions, and little ability to maintain it.
Derived from
Another map file:
An odd map file:
Dual-language Montreal wants labels in French and English, but a translation would only need one language. The file also has embedded raster graphics.
Other maps
Compare to some internally translated SVG files. Sadly, these files are generated by hand. Pulling them into an SVG editor can strip the translations.
Here's a hand-edited file that has included translations; there are 49 different IETF language tags in the file. The file also uses some SVG context sensitive features that are lost in the PNG translation.
How does the wiki access different versions? It tacks on &lang=IETF
to the link.
For example, the line
[[file:First Ionization Energy.svg|lang=de|thumb|left|German version]]
Inserts this image:
by inserting the following HTML:
<div class="thumb tleft">
<div class="thumbinner" style="width:222px;"><a href="/w/index.php?title=File:First_Ionization_Energy.svg&lang=de" class="image"><img alt="" src="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langde-220px-First_Ionization_Energy.svg.png" width="220" height="92" class="thumbimage" srcset="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langde-330px-First_Ionization_Energy.svg.png 1.5x, //upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langde-440px-First_Ionization_Energy.svg.png 2x" data-file-width="1200" data-file-height="500" /></a>
<div class="thumbcaption">
<div class="magnify"><a href="/wiki/File:First_Ionization_Energy.svg" class="internal" title="Enlarge"></a></div>
German version</div>
</div>
</div>
The anchor element makes a click go to the file page with the appropriate language selected for the SVG.
The img
element displays the picture.
Not sure what 1/1d means, but it may be a cache issue.
Image directory has lang(IETF langtag)-(size)px-(name).png
made by renderer.
The srcset
attribute provides magnify/"enlarge" options?
Notice data-file-width
and data-file-height
attributes that may anticipate problems with SVG files containing hard widths and heights.
Check langtag issue at Phab:T154237.
librsvg
does not handle hyphenated langtags correctly, so zh-hans
and zh-hant
just match to zh
and look the same:
In September 2017, I noticed the above two images (zh-hans an zh-hant) were not working at all (not just selecting the wrong version). Try just zh:
New development of Phab:T154132.
So first string works, but second does not.
src="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langzh-220px-First_Ionization_Energy.svg.png"
src="//upload.wikimedia.org/wikipedia/commons/thumb/1/1d/First_Ionization_Energy.svg/langzh-hans-220px-First_Ionization_Energy.svg.png"
Guessing that parameter extraction is now confused by hyphenated langtags.
Lazy evaluation suggest the URL is grabbed at .../thumb/...
and then parsed
Hypenated language tag problem fixe 26 October 2017 by Gilles. UI problems still present.
Want each wiki to automatically use the language tag. Somewhere there was a comment that this will come.
The periodic table just needs translations for "group" and "period".
Here's a file with just 6 sentences to translate.
Interesting problem with language variation (accent variation between pt-BR and pt-PT: Hidrogênio and Hidrogénio):
Also look forward to direct insertion of SVG files rather than the current convert to PNG.
Here's a JPEG file that is used on many projects and that people want translated. An SVG version without English captions is also desired.
A simpler diagram
Needing conversion; spot labels
Several images use numbers, but have translations on their description pages.
file not needing translation (but conversion to SVG)
2019 Wishlist survey to add Chemical File Format
Lots of diagrams with minimal text changes.
This file has some embedded fonts.... Complicated shading.
i:pgf
element) added
i:pgf
element: https://stackoverflow.com/questions/6745952/creating-illustrator-top-layers-in-svgGood targets are simple SVG files that are used on several foreign WPs. If the file isn't used, translation is not a priority.
May want to use ((translation possible))
|switch=value
for many other uses. ITS tags, for example.
Matching numbers and units (e.g., 30 volts).
Good candidates (see Commons:Category:Translation possible - SVG (switch))
Dubious SVG:
Switch-translated files that don't get much use. Note temperature scale/labeling issues.
Some files can be problematic even if converted to SVG (e.g., dense diagram, fr translations overlong)
Some files do not need translation because they do not use words. Many diagrams use numbers to identity components.
A benefit to this method is the diagram is universal. It also does not have problems with multi-line labels.
The disadvantage is the translation must be done each time the image is used: the caption must explain what the numbers mean. In other words, the image does not stand by itself but rather requires an external explanation.
In this method, an existing file is edited by replacing the existing text with translations and then saved under a new name. It is the default for Wikimedia Commons because it is easy to do. Pull the existing file into Inkscape, edit it, and save it.
The problem is it makes many copies of essentially the same file. If someone wants to change the diagram a little or change the color scheme, it must be done for each copy.
A better method uses one copy of the file. There are two methods: switch and XLIFF. Wikimedia Commons is not set up for XLIFF.
Better idea about how Commons approaches the translation issue.
The translation markup.
In an SVG file, find text blocks or spans that permit translation.
<text x="100" y="200">House</text>
converts to
<g transform="translate(100,200)">
<switch>
<text systemLanguage="de">Haus</text>
<text>House</text>
</switch>
</g>
or even
<switch transform="translate(100,200)">
<text systemLanguage="de">Haus</text>
<text>House</text>
</switch>
The addition of switch
has a small impact (roughly 16 bytes). Each clause costs at least 33 bytes plus the translation bytes. Estimate the translation is 17 bytes. Then each language adds 50 bytes. With fifty languages, a single term adds 2.5 kB. Forty terms would add 100 kB to an SVG file. That relies on the position information being hoisted so it is not repeated for each translation. File:Map of USA with state names.svg paid 400 kB even though it has hoisted location information; I don't know where the penalty is; 160 bytes per language term seems high.
For ordinary SVG files, the systemLanguage is compared to the user's preferences. SVG 1.1 has slightly odd semantics for choosing a preference, but that is being straightened out in SVG 2.0.
WP throws a further wrench in the preference calculation because it does not serve the SVG but rather a bitmap. That bitmap is built by rsvg
which only takes a single language. The lang argument should be a preference list. If the illustration does not have a desired language, then "en" may not be the best fallback. An "sr-Cyrl" might better fallback to "sr-Latn" and after that "ru". It's a minor point, and fixing it might be too much trouble. Serving the switch SVG would fix it. A German reading the de.WP might want the fallback to be "en", but a Swiss might want the fallbacks to be "fr" and "it" before "en".
A problem with serving SVG is there are some very piggish SVG files. A 200 px PNG thumbnail has a bounded size.
Consider problems when the text has leaders that point to places in the image.
Problem with expansion. Some languages have succinct labels, but others require more space. Don't consider problem when line breaks are needed. That really is an XLIFF issue.
There are things that should not be translated. Some phrases are intended to stay in a specific language; a diagram that explains translation, for example, would keep the native language. Names are often not translated (Casablanca does not become White House), but they may be Anglicized (e.g., Чебышёв becomes Chebyshev). Chemical symbols are not translated. Symbols may change (input voltage V becomes input U). What about the subscripts?
Conjugation and declension. Would not expect a dictionary to have "red house" or "three houses". Google Translate and some matching?
Display multiple outlines -- enough room so everything fits. This could make diagrams in some languages look poor.
its:term
and friendsThere is an issue with mixins and validation.
There is an issue with <title>
, <desc>
, and <hint>
elements because they are not switched. Conditional processing applies to display elements, but not others such as titles and descriptions. Consequently, we cannot do this:
<svg ...>
<switch>
<title systemLanguage="de">Wien</title>
<title systemLanguage="en">Vienna</title>
<title>Vienna</title>
</switch>
</svg>
That means it will be difficult or impossible to translate titles/tooltips. On the other hand, XLIFF would have no trouble with those elements.
SVG 2.0 working draft recognizes this problem,[1] but does not use systemLanguage. Instead, title
and desc
may appear multiple times with different lang
attributes (which are single valued). The construct (which is not yet supported in browsers: Issue 67) is
<svg lang=en...>
<title>Vienna</title>
<title lang="de">Wien</title>
<title lang="ru">Вена</title>
...
</svg>
More exotic issues: mouseover. I believe this fails and is better handled with XLIFF or separate rectangles.
<rect x="100" y="200" w="20" h="40">
<title>House</title>
<title lang="de">Haus</title>
<desc>Blue</desc>
<desc lang="de">Blau</desc>
</g>
Several files do not keep the translations together but rather separate languages into a planar configuration. Consequently, there is only one switch
element and only one clause per language. Not repeating the language selection clausest suggests a byte savings, but each text
element must repeat the position information — something whose cost would match the cost of repeating systemLanguage="xx"
. In addition, the planar configuration does not keep the translated terms together. Any equivalence must be inferred from the similarity in position in the diagram or order within the clauses.
<switch>
<g systemLanguage="de">
<text x="100" y="100">Strasse</text>
<text x="100" y="200">Haus</text>
</g>
<g>
<text x="100" y="100">Street</text>
<text x="100" y="200">House</text>
</switch>
In addition, there are different methods of planar association. A plane might be a g
element with text
children, or it might be a text
element with tspan
children. The latter method complicates multiline phrases. If "Atlantic Ocean" needs to be broken into two lines, it is better if the lines share a unique parent so they hang together. A simple view is each phrase should exist in its own text
element; if the phrase is broken over lines, then those lines are tspan
elements within the text
element. Otherwise, all the words are just shuffled together.
I would deprecate the planar method, but Graphics Lab editors seem to favor it.
Here's a switch file that has tspan
elements to do multiline translations.
If userLang=ec, we take the first. If userLang=ec-EL, we bypass the first.
Many SVG files have language order problems. (Is there a allowReorder
section?) For example, this ordering is wrong because SVG 1.1 rules hide the following sr-EC and sr-EL langtags:
<text systemLanguage="sr">Актиноид</text>
<text systemLanguage="sr-EC">Актиноид</text>
<text systemLanguage="sr-EL">Aktinoid</text>
A better order is
<text systemLanguage="sr-EC">Актиноид</text>
<text systemLanguage="sr-EL">Aktinoid</text>
The reorderings have slight problems because "sr-EC;q=1.0,sr-EL;q=0.8,sr;q=0.5" and "sr-EL;q=1.0,sr-EC;q=0.8,sr;q=0.5" will always choose the first option under SVG 1.1 rules.... Maybe the proper viewpoint is the undesired script must be set below the "sr" q value.
Reasonable sorting algorithm? Since systemLanguage
can take many langtags, a reordering cannot solve the problem in general.
Here's a single-switch/multi-switch file that put Ukrainian translation after the default translation (uk translation now reordered). Fault line was incorrect (should be fault plane). Only a few terms to translate: epicenter, fault plane, hypocenter, and focus. Wiktionary gives good coverage for British epicenter, modest for focus, and poor for the rest. File has funky rotation using a transform matrix despite line drawn elsewhere.
A file with no default language:
SVG renderers may not support all SVG features. For example, rsvg
does not support textpath
; consequently, many users take the text and convert it to curves.
Switch features (requiredFeatures
attribute and hasFeature
DOM)
The feature#Text
should signal that textpath
is not available. rsvg
does not claim feature#Text
:
However, some browsers may claim the feature but not do all the alignments; that may be inappropriate.
Does rsvg have partial support?
<switch>
<textpath requiredFeatures="http://www.w3.org/TR/SVG11/feature#Text" ...> ... </textpath>
<path ...> ... text as curves ... </path>
</switch>
There's also the text baseline alignment mess to figure out. dy
can be used, but there are also some CSS modifications that can be done. It may be that CSS baseline alignment works when baseline not available, but it seems misused.
There's also SVG 1.2 textArea
with its feature:
SVG Optimizer additions?
Mediawiki blocks some elements....
Trying direct SVG inclusion (blocked):
<svg xmlns="http://www.w3.org/2000/svg" width="320" height="160"> <rect x="0" y="0" width="320" height="160" fill="cyan" /> </svg>
Using tag magick worter...
<svg xmlns="http://www.w3.org/2000/svg" width="320" height="160"><rect x="0" y="0" width="320" height="160"></rect></svg>
Try object:
<object width="320" height="160" data="https://upload.wikimedia.org/wikipedia/commons/1/1d/First_Ionization_Energy.svg" type="image/svg+xml" > </object>
Trying bdi: bdi and bdo. These elements work.
User could supply translations as in the current SVG Translate tool.
We don't get a lot of languages at once. Want the image to explode. User draws and SVG diagram and labels it. Diagram then becomes available with translations in many languages.
Good coverage.
Get a lot of languages, but theres a problem with getting the wrong sense of a word.
I don't know how to qualify the field to Google Translate, and the translations are probably better for long sentences rather than isolated words.
Just make it easy to correct the translations.
Conventional approach is a database of translations.
This could have field and tracked corrections. Require some subject indicator or word sense.
There's a difficult issue here. Image must be accurately labeled, but then the labels may be too fine (no translations) or too broad (word confusion still exists).
The field could be gleaned from some categories for the image.
That brings up a separate issue. Is the Gibraltar map present many times in a category? Better to have one image and select the right language depending on the user's preference.
Articles have interlanguage links, and they can be used to extract translations.
The image File:BirdBeaksA.svg has some interesting issues. It is not about the actual birds but rather their beaks. The Commons page also has interesting links.
Maps often represent a point in time.
So Q item current versus another time.
Chennai: Chennai (Q1352), Madras in aliases: Madras, official name (P1448) : Madras
Burma, Myanmar Myanmar (Q836), aliases: Burma, Republic of the Union of Myanmar, 🇲🇲, Union of Burma, MM, BUR, MYA, official name (P1448): Republic of the Union of Myanmar
St Petersberg, Petrograd, Leningrad. country (P17) changes a lot.
Planet Nine search has many orbital maps in PNG. It's easy to draw an ellipse in SVG, but the coordinates need to be mapped. The Geo mapping probably does not cover that issue well. Project into the ecliptic.
Wikidata should provide the orbital elements, but that has not made it to Wikidata yet. See (496315) 2013 GP136 (Q22231120), which has no orbital elements. Compare to (496315) 2013 GP136 which has the elements.
Wikidata could be a star catalog. Cygnus (constellation).
If I know the Q number for a list (eq (Q42)), then (format= json, jsonfm, xml, xmlfm)
Look at some items.
Failed to render property P2237: P2237 property not found.
(property of property) P558 unit symbolFailed to render property P558: P558 property not found.
electronvolt (Q83327)Problems...
This gives several Qs
API definition
An object may have a WD item but its components may not.
Consider an illustration of a MOSFET. It may have drain, gate, source, channel, substrate, insulator, depletion region.
See illustrations at
Another series (with damaged illustrations)
Wikidata hierarchy does not have components.
Do have
Consider Wictionary....
Had trouble with earlier, but starting to get results:
SELECT ?item ?itemLabel ?symLabel {
# [] means I do not care what the value is
# ?item p:P246 [] .
?item wdt:P246 ?sym .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .
}
}
# instances of chemical elements
SELECT ?item ?itemLabel ?number {
# instance of a chemical element
?item wdt:P31 wd:Q11344 .
# get the atomic number
?item wdt:P1086 ?number .
# do not include the really high numbers
filter (?number <= 118) .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .
}
} ORDER BY ?number
SELECT ?item WHERE {
?item wdt:P31 wd:Q11344 .
filter not exists { ?item p:P155 [] }
}
syntaxhighlight
SELECT ?item WHERE {
?item wdt:P31 wd:Q11344 .
filter not exists { ?item p:P155 [] }
}
href
.((SPARQL))
:
SELECT ?item WHERE {
?item wdt:P31 wd:Q11344 .
filter not exists { ?item p:P155 [] }
}
#
SELECT ?item ?itemLabel ?mass ?atomnumber ?pred ?predLabel
{
?item wdt:P31 wd:Q11344 ;
p:P31 ?stmt ;
wdt:P1086 ?atomnumber . # data type is a quantity
?stmt pq:P155 ?pred .
# not all elements have mass!
filter not exists { ?item wdt:P2067 ?mass . } .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .
}
} ORDER BY ?atomnumber
# comment
SELECT ?item ?itemLabel ?mass ?atomnumber
{
?item wdt:P31 wd:Q11344 ;
wdt:P1086 ?atomnumber . # data type is a quantity
# not all elements have mass!
filter not exists { ?item wdt:P2067 ?mass . } .
filter (?atomnumber <= 118) .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .}
} ORDER BY ?atomnumber
# comment
SELECT ?item ?itemLabel ?ie ?unitLabel ?ord ?atomnumber
{
?item wdt:P31 wd:Q11344 ;
wdt:P1086 ?atomnumber .
# not all elements have ionization energy
{ ?item p:P2260 ?stmt .
?stmt pq:P1545 ?ord .
?stmt psv:P2260/wikibase:quantityAmount ?ie .
?stmt psv:P2260/wikibase:quantityUnit ?unit .
} .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .}
} ORDER BY ?atomnumber
I added mass for hydrogen, lithium, and nitrogen.
Push endpoint:
The wikidata entry for oxygen oxygen (Q629) gave O2 as an alias; I removed it because the article is about the element. There is a dioxygen (Q5203615) which is a simple substance but is not an instance of a gas. fluorine (Q650) is an instance of a gas. nitrogen (Q627) is not an instance of a gas. carbon (Q623) is not an instance of a solid. There's a logical problem here. An element is an element; it does not have to be in any phase. I need to distinguish state of matter (Q11430) and phase (Q104837), an aspect of state. Some elements form diatomic molecules, and those molecules can be solid, liquid, gas, plasma under certain conditions. Oxygen can be, for example, a liquid. Any statement about state/phase should contain conditions. But can just an element be something if it will form molecules? Copper copper (Q753)???
On 21 August 2017, Wikidata sez there are 1,508 United States Senators who are still in office (only 1,504 if we insist the Senators be human; Wikidata knows 1,319 are dead):
# find US senators
SELECT ?person ?personLabel ?stateLabel ?bdate ?ddate ?endtime
{
# BIND (wd:Q121337973 AS ?person) .
# want a human
?person wdt:P31 wd:Q5 .
# person held a position
?person p:P39 ?stmt .
# that position is US senator
?stmt ps:P39 wd:Q4416090 .
# position has not ended
filter not exists { ?stmt pq:P582 ?endtime } .
# grab the electoral district if it exists
optional { ?stmt pq:P768 ?state . }
# grab a birth date if it exists
optional { ?person wdt:P569 ?bdate . }
# grab death date if it exists
optional { ?person wdt:P570 ?ddate . }
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .
}
} ORDER BY ?stateLabel
I did this query and spent a day pushing down to 100 entries:
# work...
SELECT ?person ?personLabel ?stateLabel ?bdate ?sdate ?ddate
{
?person p:P39 ?stmt .
?person wdt:P569 ?bdate .
?person wdt:P31 wd:Q5 .
?stmt ps:P39 wd:Q4416090 .
optional { ?stmt pq:P768 ?state . } .
optional { ?stmt pq:P580 ?sdate . } .
filter not exists { ?person wdt:P570 ?ddate . } .
filter not exists { ?stmt pq:P582 ?endtime . } .
filter not exists { ?stmt pq:P1366 ?succby . } .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .} .
} ORDER BY ?stateLabel
I ran this query with several governor-of-state. VP Pence was still listed as governor of Indiana.
# comment
SELECT ?item ?itemLabel ?ord ?startdate ?enddate ?successorLabel WHERE {
?item p:P39 ?stmt.
?stmt ps:P39 wd:Q16147601.
OPTIONAL { ?stmt pq:P580 ?startdate. }
OPTIONAL { ?stmt pq:P582 ?enddate. }
OPTIONAL { ?stmt pq:P1366 ?successor. }
OPTIONAL { ?stmt pq:P1545 ?ord. }
SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
ORDER BY ?startdate
Click here to launch the Wikidata query
# comment
SELECT ?item ?itemLabel ?headLabel ?govLabel WHERE {
?item wdt:P31 wd:Q35657.
optional { ?item wdt:P1313 ?head } .
optional { ?item wdt:P6 ?gov } .
SERVICE wikibase:label { bd:serviceParam wikibase:language "en". }
}
ORDER BY ?itemLabel
Click here to launch the Wikidata query
Fixed list of California Senators.
# comment
SELECT ?person ?personLabel ?stateLabel ?bdate ?sdate ?endtime ?ddate
{
?person p:P39 ?stmt .
?person wdt:P569 ?bdate .
?person wdt:P31 wd:Q5 .
?stmt ps:P39 wd:Q13217683 .
{ ?stmt pq:P768 wd:Q99 . } . # ?state . } .
{ ?stmt pq:P580 ?sdate . } .
OPTIONAL { ?person wdt:P570 ?ddate . } .
OPTIONAL { ?stmt pq:P582 ?endtime . } .
# filter not exists { ?stmt pq:P1366 ?succby . } .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .} .
} ORDER BY ?sdate
Click here to launch the Wikidata query
Different languages use different scripts, so the symbols used for those units may vary.
# comment
SELECT ?item ?itemLabel ?unitLabel ?unitLang ?scriptLabel ?langLabel
{
# coulomb eV second
VALUES ?item {wd:Q25406 wd:Q83327 wd:Q11574}
# find an item with a unit symbol
# ?item p:P558 ?stmt .
# unit symbol is now 5061
?item p:P5061 ?stmt .
# get the unit symbole for each language
# P5061 has type of monolingual text.
# monolingal text has a mandatory language
?stmt ps:P5061 ?unit.
# bind so we can report it in results
BIND (lang(?unit) AS ?unitLang) .
# filter (lang(?unit) = "zh-hant") .
# filter (langMatches(lang(?unit), "zh")) .
# this no longer works
# get the script qualifier for the statement
optional { ?stmt pq:P282 ?script . } .
# this no longer works
# get the language qualifier for the statement
optional { ?stmt pq:P407 ?lang . } .
SERVICE wikibase:label {
bd:serviceParam wikibase:language "en" .}
} ORDER BY ?itemLabel
Click here to launch the Wikidata query
Songs should have a composer and a lyricist. (Compare instrumental Classical Gas (Q489275).)
John Lennon (Q1203) and Paul McCartney (Q2599) had a contract to be joint. Look at all songs performer (P175) The Beatles (Q1299). That should give an interesting result with respect to RDF bag or individual composer (P86) and lyricist (P676). Some songs get hits for Lennon–McCartney (Q239177).
# comment
SELECT ?song ?songLabel ?composer ?composerLabel ?lyricist ?lyricistLabel WHERE
{
# find an instance of a song
# ?song wdt:P31/wdt:P279* wd:Q7366 .
# instance of a musical composition
?song wdt:P31 wd:Q105543609 .
# form of creative work is a song
?song wdt:P7937/wdt:P279* wd:Q7366 .
# the performer is The Beatles
?song wdt:P175 wd:Q1299 .
# get the composer
OPTIONAL { ?song wdt:P86 ?composer . }
# get the lyricist
OPTIONAL { ?song wdt:P676 ?lyricist . }
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
} ORDER BY ?songLabel
Click here to launch the Wikidata query
# comment
SELECT ?typeface ?typefaceLabel
WHERE
{
?typeface wdt:P31/wdt:P279* wd:Q17451 .
SERVICE wikibase:label { bd:serviceParam wikibase:language "[AUTO_LANGUAGE],en". }
}
ORDER BY ?typefaceLabel
Click here to launch the Wikidata query
Diagrams are often associated with particular fields of study such as chemistry, physics, or mechanical engineering.
The Wiktionary translations are spotty (especially with technical terms). Gloss/distinctions are troublesome or even missing; compare wikt:stripper. Some common terms have good coverage. See, for example, wikt:radon, but notice Uzbek Latin and Cyrillic entries. Each entry is a span
with class
attribute of "Latn" or "Cyrl" and lang
attribute of "uz".
In 2019 Wishlist Survey, there's a proposal to link Wikidata items to Wiktionary defs
In 2019 Wishlist Survey, there's a proposal to rationalize Wiktionary translations.
For reference:
Mechanical engineering
Diagram (which has been removed from Commons) has only one broken word....
English | German |
---|---|
deep drawing | Tiefziehen (deep pull) |
workpiece | Werkstück |
blank | Ronde (not a place on a form) |
blank holder | Niederhalter (hold down) |
die (piercing, blanking, drawing) | Ziehmatrize (pull die) |
drawing die | Ziehmatrize |
punch | Ziehstempel (pull stamp) (not what a boxer does) |
stop | |
automatic stop | |
clearance | (not the sale) distance between two moving objects [sic] |
die holder | |
stripper | Abstreifer (=wiper/scraper) (not the entertainer) one who removes their clothes in a sexually provocative manner chemical or tool to remove paint, sheathing tool used to strip tubing |
guide pins | (plural!) |
Physics
Diagram has lots of multiline statements. A better layout would use single-line labels, but it would look different. There is a mostly single-line version.
English | German |
---|---|
atomic number | Ordnungszahl Kernladungszahl Atomnummer |
glow discharge | Glimmentladung |
cathode fall | (missing) |
Cathode | Kathode https://de.wiktionary.org/wiki/Kathode |
Aston dark space | astonscher Dunkelraum (m) |
cathode glow | Kathodenglimmhaut |
cathode dark space Crookes dark space Hittorf dark space (sp?) |
hittorfscher Dunkelraum (m) Crookes Cathode |
negative glow | negatives Glimmlicht (n) |
Faraday dark space | faradayscher Dunkelraum (m) |
positive column | positive Säule (f) |
anode glow | Anoden-Glimmlicht anodisches Glimmlicht (n) |
anode dark space | anoden-Dunkelraum anodischer Dunkelraum (m) |
anode | Anode |
Unfortunately, Wiktionary does not have good coverage of the terms. I had put many of the physics terms in some time ago.
Would want a program that updates a diagram. If new translations are added, then they should be incorporated.
For UI, what hacks can I do with DOM?
Watt makes many points.
style=
because that confuses content and presentation:
style
attribute, and something that should be avoided. Contrary to popular belief the style
attribute (note, that's attribute, not element) doesn't separate content from presentation, and unless you need to override properties set by a CSS selector, it is best to use the SVG formatting attributes instead.Hey, there is a translation in the file namespace:
Which should relate to
Where srt is SubRip format. HTML 5 track element wants WebVTT (vtt). The translation might be faster to wait for .srt to change and then convert it to .vtt.
track
information:<track src="//commons.wikimedia.org/w/index.php?title=TimedText:The_GLAM-Wiki_Revolution.webm.de.srt&action=raw&ctype=text%2Fx-srt" kind="subtitles" type="text/x-srt" srclang="de" label="Deutsch (de) subtitles" data-mwtitle="TimedText:The_GLAM-Wiki_Revolution.webm.de.srt" data-mwprovider="local" data-dir="ltr"/>
text/x-srt
. Somewhere there was a note that track
required Content-Type
.How are these set up?
Issues...
Figure out MyLanguage hack. Is resolved during translation?
There is an existing tool to aid SVG translations. See
The pages states, "This tool has been used to translate approximately 865 files since March 2011."
Just realized a confusing issue with this tool. X creates the original en version of an image. Y translates the image into fr; the Commons page suggests translating the file. Z comes along and translates fr version into de. We've translated a copy. Want that functionality (some translators won't know en), but the non-translated source should come from the original rather than the copy.
Apparently some of the translated files have been translated by pulling them into Inkscape and just editing the labels.
SVG Translate is a PHP server-side tool.
It offers no translation help.
It throws up the text/tspan contents and asks the user to translate them.
Instead of using DOM processing, it uses pattern matching. The algorithm acknowledges that there will be problems for unusual element names. Stuffing a switch-translated SVG should have interesting results.
There is no attention to its:
rules and attributes.
It copies the file and edits it. Consequently, there are now multiple copies (one per language) to maintain of the essentially the same information.
The industry route would be to have a skeleton file with XLIFF files to fill in the translations. Wiki are not set up to do that. There is an i18n file hierarchy for templates.
When I tried to use it, it threw an exception.
Sometime later (ca. September 2016) it would throw an exception about Intuition::getLangNames()
testing File:Planetary transit.svg on https://commons.wikimedia.org/w/api.php Fatal error: Call to protected method Intuition::getLangNames() from context 'SVGtranslate' in /data/project/svgtranslate/public_html/svgtranslate.php on line 395 Call Stack: 0.0039 675248 1. {main}() /data/project/svgtranslate/public_html/index.php:0 0.1270 3318376 2. SVGtranslate->do_step() /data/project/svgtranslate/public_html/index.php:57 0.1270 3318512 3. SVGtranslate->generate_second_form() /data/project/svgtranslate/public_html/svgtranslate.php:781
OK, assuming svgtranslate from
is the beast, then
$html .= '<tr><th align="right">' . _html( 'th-language' ) . _g( 'colon-separator' ) . '</th><td>';
$html .= "<select name=\"targetlanguage\" style=\"width: 40em\">\n";
$langnames = $I18N->getLangNames();
$default = $I18N->getLang();
foreach( $langnames as $code => $name ){
$html .= "\t<option value=\"$code\"";
if( $code == $default ){
$html .= " selected='selected'";
}
$html .= ">$name</option>\n";
}
Intuition has published interfaces getLangName
and getLangNames
.
An edit in March changed public getLangNames
to protected getLangNames
on line 659 (now 661).
Code located in
Got phabricator wiki account and left comment about origin of SVG Translate bug.
Need GitHub account to complain about Intuition (but there may be a fork?).
I left some mail, and Krinkle fixed Intuition to make interface public again.
Well, at least more parts are working. I never tried to save anything, but there is a report that saving the file does not work.
its:translate="no"
).Sadly, the source JPEG gets more use on xx.WPs:
It is a standard SVG map that has many translations, but it has some translation issues.
It does not specify a lang
in its svg
element.
SVG does not validate. Issues Inkscape and RDF warnings. Then 4 errors for illegal characters in id string such as:
id="texte "Pays""
The problem is the normalized value of the xml:id
should be an NCName, and NCNames should not have spaces or quotation marks. See http://www.w3.org/TR/xml-names11/#NT-NCName
Much of the file is graphics (extensive paths for borders) and internal bookkeeping (including worthless perspective defs); little is the actual text. It might be interesting to run svgo
on it. Does svgo
dissect the style
attribute? Yes, see convertStyleToAttrs
; moveElemsAttrsToGroup
.
All capital names such as "SPAIN", "FRANCE", "ENGLAND", and "ITALY" make translation harder.
Multi-line labels such as "HOLY ROMAN EMPIRE" (Holy Roman Empire (Q12548)), "ATLANTIC OCEAN" (Atlantic Ocean (Q97)), "Mediterranean Sea" (Mediterranean Sea (Q4918)), and "Dutchy of the English King". Looking at SVG source (with lots of Inkscape details) shows that multiline text is done correctly with embedded tspan
elements. That may be good text line handling by Inkscape. However, the SVG Translate tool breaks them out as individual tspan
elements so the grouping is lost. Better to have "Mediterranean Sea" translated to "Mar Mediterráneo".
Stolen from Talk:Oroville Dam crisis:
Copying is a poor method of using a symbol library. Copies are difficult to maintain. In its defense, SVG xlink:
is limited.
Inkscape may get in the way with its styles.
Selecting and copying the two-line "Mar Mediterráneo" from the SVG gets "MarMediterráneo"; the copy did not add a space even though the words were on separate lines. SVG should have clipboard copy rules for text
and tspan
elements. The text
element has xml:space="preserve"
, so that should control. Newlines are supposed to be turned into spaces. However, there is not a newline within or after tspan
elements, so no space should be inserted. That would explain why no spaces occur between the separate lines.
<text
sodipodi:linespacing="125%"
id="text4976"
y="662.791"
x="126.492"
style="font-size: 32px; font-style: normal; font-variant: normal; font-weight: bold; font-stretch: normal; text-align: center; line-height: 125%; text-anchor: middle; fill: rgb(100, 100, 100); fill-opacity: 1; stroke: none; display: inline; font-family: Arial;"
xml:space="preserve"><tspan
y="662.791"
x="126.492"
sodipodi:role="line"
id="tspan4982">Duchy of</tspan><tspan
id="tspan4989"
y="702.791"
x="126.492"
sodipodi:role="line">the English</tspan><tspan
id="tspan4991"
y="742.791"
x="126.492"
sodipodi:role="line">King</tspan></text>
Including a trailing space within the first tspan
should insert the space, but it might confuse text-align="middle"
. A newline might make more sense if it just turns into a space on select and copy:
<text xml:space="preserve"><tspan>Dutchy of</tspan>
<tspan>the English</tspan>
<tspan>King</tspan></text>
The above is wrong due to "preserve"
: there would be too many spaces. It controls the initial and final newlines, but the indentation would introduce more spaces. Using xml:space="default"
would have the better semantics; it deletes newlines but merges spaces; it also deletes leading and trailing spaces.
<text xml:space="default">
<tspan>Dutchy of</tspan>
<tspan>the English</tspan>
<tspan>King</tspan>
</text>
The above might get the right result; the spaces, however, will come from the indentations. Newlines will be deleted, tabs become spaces, leading and trailing spaces are deleted, and then spaces are coalesced.
It looks like the xml:space
issue is more complicated and changing. If I try it in HTML, I cannot produce the expected problem. Many browsers and XML tools ignore xml:space
.
Looks like xml:space
is deprecated but still alive in SVG 2.
The recommendation is to use the CSS white-space property.
Another issue is whitespace in XSLT.
There's also an issue with hyphens. If a line is broken by hyphenating a word, then it would be nice if the copy removed the word break hyphen but not disturb an ordinary hyphen.
The checkered career of the soft hyphen:
Argues the semantics started out being the soft hyphen was the thing added at the end of the line by the line breaking agent, but then the semantics shifted to signal a line-breaking opportunity. Points to different standards, offers an interpretation of the standard's meaning, and covers some implementations. For example, ­ is a real hyphen in Microsoft Office (try cut and paste!); Office uses something else to signal line breaking opportunities; when Office spits out HTML, its optional hyphen is converted to a ­.
-ms-hyphens:auto
; Chrome only on Android and Mac.Text selection:
There's some indication that limited amounts of text may be selected. It may be confined to a text
element.
Back to the map example.
IETF langtags handed as native language local only; langtags not shown, but it looks like the list is in langtag order.
Resulting labels took effort. Some label layouts have been tweaked.
After diagram was made, somebody added additional cities for time of Joan of Arc. Possible that previous translations did not get the benefit of the additions (see -sv version).
Color scheme is poor, but now exists in several instances. It would be better to choose a good color scheme in one place.
The map also shows another problem: capitalization. The country is "Spain", but the map text element uses "SPAIN". That complicates string matching. It would be better to use "Spain" and "England" as the actual text (rather than "SPAIN" and "ENGLAND"), set a class "country", and then use CSS style to uppercase the presentation text:
<style type="text/css" media="media types" title="advisory title" >
<![CDATA[
text.country: {text-transform: uppercase; }
text.ocean: {text-transform: uppercase;
font-style: italic;
color: blue; }
text.sea: {font-style: italic;
color: blue; }
]]>
</style>
Interestingly, CSS styles would do their proper role here: they affect the presentation while keeping ordinary spellings ("Spain") would preserve content. The class information could be used as additional context in the translation process.
Possible text-transform values are none
, uppercase
, lowercase
, capitalize
, full-width
. In addition, a CSS property can be inheirit
or initial
, but initial
is not cap first letter of the string but rather the default (initial) value.
Also, capitalization problems arise elsewhere. For an earthquake diagram, it would be better to use "epicenter" rather than "Epicenter" to aid translation; a capitalize transform would then capitalize the word. It would be sticky for "center of the universe" becoming "Center Of The Universe". It may also have bizarre consequences if other languages do not have same notion of capitalization. It might be better to just avoid the capitalization completely and leave the picture with "epicenter" rather than "Epicenter".
Plurals raise the same problem. The chemical elements include a class of noble gases, and there is some variation in how to refer to the class or membership in the class. A diagram might identify neon as a noble gas; the wikidata class is noble gases (Q19609) (singular). In contrast, wikidata uses the plural for alkaline earth metal (Q19563) with an alias for "alkaline earth metals". Aliases do not cover the missing plural/singular. The issue gets much more complicated for languages that decline adjectives: the string distance increases.
A map spaced letters out: "ATLANTIC OCEAN" became "A T L A N T I C O C E A N". The spacing can be done with the letter-spacing
property. Can apply capitalization and spacing at the same time with style="text-transform:uppercase;letter-spacing:0.5em"
:
That opens up a whole area of single-language edits (no translation skills required) to a diagram or map to improve i18n workflow. Editors who are not translators can edit diagrams to be translatable.
Random test of abbrev:
TT
My understanding is SVG Translate generates a new SVG file from an existing one; it uses neither XLIFF nor switch
. It was written some time ago by user:Nikola_Smolenski, but he had moved on ca 2010, so there was a request for a new programmer for continued maintenance.
User:Jarry1250 took over some maintenance; User:Luxo and User:Leyo chimed in. SVG Translate sources are on github, but they haven't been updated in a year and a half.
SVG Translate is clearly a significant effort. In browsing some sources, it has a multilingual user interface. Such an interface would be needed for editors who only speak their language. It also did some rights management. I have not found significant documentation for either the user or the program architecture.
DerivativeFX may be a licensing tool by Luxo. http://toolserver.org/~luxo/derivativeFX/deri1.php
Discussion about DerivativeFX:
SVG Translate apparently does not use DOM but rather pattern matching. Test for SVG file looks at the file extension rather than the file.
When I tried to run it recently, it threw an error at line 395. An array of all languages is being released when that array is protected. Somewhere I had found a bug report.
Looking at the bad case selection bug (eg zh-Hans). The page HTML has
<form action="/w/index.php">
Render this image in
<select id="mw-imglangselector" name="lang">
<option value="bs">Bosnian (bs)</option>
<option value="cs">Czech (cs)</option>
<option value="de">German (de)</option>
<option value="en" selected="">English (en)</option>
<option value="fi">Finnish (fi)</option>
<option value="hr">Croatian (hr)</option>
<option value="sh">Serbo-Croatian (sh)</option>
<option value="sr">Serbian (sr)</option>
<option value="sr-EC">Serbian (Cyrillic script) (sr-EC)</option>
<option value="sr-EL">Serbian (Latin script) (sr-EL)</option>
<option value="vi">Vietnamese (vi)</option>
<option value="zh-Hans">Simplified Chinese (zh-Hans)</option>
<option value="zh-Hant">Traditional Chinese (zh-Hant)</option>
<option value="zh-cmn-Hans">zh-cmn-Hans</option>
<option value="zh-cmn-Hant">zh-cmn-Hant</option>
</select>.
<input type="submit" value="Go"/>
<input type="hidden" value="File:First_Ionization_Energy.svg" name="title"/>
</form>
So clicking the submit button leads to
Apparently the problem is the option value="zh-Hans"
should be lowercased when page is generated, or index.php
should lowercase the lang
parameter.
The problem is more widespread because mixed-case languages are not properly handled:
Different from SVGTranslate
. Wants to use switch
:
There's a comment about standard form.
Could not find it, but there is IsTranslationReady()
(switch
installed).
Recent translation pushes, but not much in development.
Find documentation.
Does translation control get in the way?
Burt; Jarry1250.
Other players. Rilke. Perhelion. Nikerabbit. Sameboat.
Wishlist... link to other parts
The apparent goal is this approach is to mark sections of wikitext as translatable while shielding other markup (such as tables and expressions). A <translate>
element is wrapped around material that should be translated. In ITS terminology, that element is a span
with attribute its:translate="yes"
. It also has a goal of breaking text to translate into smaller chuncks.
I don't see any notion of category or hint, but it may be there.
I do not see any notion of its:term
, but it may be there. I still am not certain whether the publication "/tech/news" should be treated as a name (and not translated) or an idea that should be.
The activity seems to be conservative: a translation and a review. Apparently nothing is changed until the translation is made.
More
JQuery.i18n
Language attribute
<bdi>
: https://developer.microsoft.com/en-us/microsoft-edge/platform/issues/10954071/
dir=
implies isolation.<bdi>
, dir
defaults to auto
<bdo>
, one must specify dir=lrt|rtl
There are some bidirectional text (BIDI) issues with Unicode that I need to sort out. Unicode has some default processing, and sometimes that default can cause confusion. There are some fences that can be constructed to avoid problems. I need to find examples that display the issues.
Consider an SVG diagram that wants to href="some wiki article"
. The problem is the URI wants to point to the article for the appropriate language: en:Radium or ru:Радий. That can be done with switch
statements that give the alternate spelling and the appropriate URI.
If just the chemical symbol is given (Ra), then there is no need to give the alternate spellings (to push the point).
Wiki-independent redirect: https://foo/mumble?wiki=en?article=Radium -> redirects to the AcceptLanguage version of en:Radium.
Radium is radium (Q1128).
Instead of AcceptLanguage, CSS could grab the lang
variable.
How does Serbian work? Are there Latn and Cyrl versions?
Bad handling of IETF langtags.
Translations would also be possible on SVG animations if WP served the SVG files. Animation GIFs have been translated:
Commas and hyphens prevent Edge SSML from working: