From Astronomy Buff I learned of an effort to digitize a century’s worth of astronomical plates.
Back in the day, before digital detectors, astronomers used glass plates to take images of the sky. Film is too flexible, and cannot be used to make accurate measurements of the positions of astronomical objects. Instead, companies like Kodak would spray the light-sensitive emulsion onto glass plates, which could be loaded into a large chamber on the back end of a telescope. The plates were big; 5×7 inches, 8×10 inches, and some were monsters, 18 inches on a side or more.
Using plates presented many problems. One is, they were heavy. Another was, duh, they were glass, and tended to break easily. A third one, weird as it may sound, is knowing which side had the emulsion on it! You can’t look — they’re light sensitive! All this has to be done in the dark, in a dark room. If you rub the plates, you mess up the emulsion and leave oil from your fingers. One classic way is to taste a corner! The emulsion has an odd, flat taste. Another way (invented by me — or at least I came up with it myself in grad school when we used plates as part of an imaging class) was to tap a corner of the plate on your front teeth. The glass side went tink! and the emulsion side went tunk! That worked every time.
A big problem is that plates aren’t very sensitive to light. They absorb only about 2-5% of the light that hits them, as opposed to modern digital detectors, which can detect well over 95% of the light that hits them. So you might think that old plates are worthless, right? We can just retake the data with better cameras now!
Bzzzzt. Nope. The sky changes. Galaxies change brightness, stars move (slowly, but some plates are 100 years old), supernova remnants expand. Old astronomical plates are literally a priceless, irreplaceable treasure.
And there lies the big problem. The Harvard Observatory has a collection of something like 500,000 plates — yes, half a million. These fragile slices of the past are hard to store. Imagine just how much they weigh in total! So the Observatory is embarking on a project to digitize all those plates.
If that sounds like a big job, well, yeah. We’re talking petabytes of data! A petabyte is a thousand terabytes, and a terabyte is a thousand gigabytes. Getting the picture (harhar)? This is millions of gigabytes of data.
Yeah. It’s a big job.
They were able to get enough money to buy the digitizer and start the project, but they’ll need a few million bucks to get the job done. If there are any very wealthy BABloggees out there, now’s the time to make yourself heard. You can pay per plates!
I was able to look at an old plate taken at the University of Virginia’s telescope when I was in grad school there. I don’t remember when it was taken; probably in the 1930s or so. It was of the globular cluster M13, a glorious beehive of stars about 25,000 light years away. Had the plate not absorbed those photons, they would have streamed on their way, and they’d be, oh, some 70 trillion kilometers past the Earth by now (actually, they would have fallen to the floor of the observatory, but give me this poetic image). That old photo represented the cluster as it was seen more than half a century ago, a view we will literally never see again.
I hope Harvard is able to complete this noble project. The Universe goes on with or without us, but we have a chance here to freeze it, if only for a moment, and see what it was like in days gone by.
July 12th, 2007 at 9:59 pm
While visiting Yerkes about ten years or so ago it was explained that the telescope’s optics had never been disassembled during its entire career of about a century. This made it rather unique in that photos taken of specific objects decades apart would be uncomplicated by any differences in the optical paths, leaving only changes due to the dynamics of the objects themselves. Thus some astronomers were at that time using the 40-incher for that exact purpose. This meant continuing to use glass plates for the photography, and Kodak was hired to manufacture a large run, which would not be repeated. As I recall, the scope was slated to be fitted with CCD imagers once the supply of glass was exhausted.
But glass is lovely. I had the chance to sift through large numbers of old plates in the vault at Lowell for a few days several years ago. It was fascinating to hold what were literally historic objects in my hands. Not only were there huge plates from the observatory’s solar eclipse expeditions, but also tons of the smaller spectrographic plates done on the Brashear instrument by Vesto Slipher, which eventually were the the first data to suggest the uniform expansion of the Universe. It was sweet.
July 12th, 2007 at 10:16 pm
Wouldn’t say a fingernail or a penny work as good as teeth? Not sure I’d want to be putting a piece of glass coated with chemicals and licked by 100 years of astronomers in my mouth.
July 12th, 2007 at 10:19 pm
Hmm now that I think about it those would have been new plates so I guess they would have been astronomer spit free and your hands were probably full holding the big piece of glass. I guess that’s why you’re the astronomer.
July 12th, 2007 at 10:45 pm
Sheesh, Phil, you didn’t catch the Bad Astronomy! From the slideshow I quote, “Astronomy, as a science, made quantum leaps forward with the advent of photography.” Quantum leaps? What the hey? They are progressing by infinitesimal leaps and bounds?
BA aside, this sound like a long, tedious process. My dad has hundreds of old slides that he wants to digitize. 50,000? Oh, oops, misread that. 500,000? Yeesh. LOOONG time spent on the scanner.
July 12th, 2007 at 11:18 pm
So here’s my question - would you, given the couple million to spend, put it towards this, or some other project, if you had to choose?
That is, in a world where every million counts, is this the best use of the money, do you think? Or given the scarcity of funding, are there better projects?
July 12th, 2007 at 11:38 pm
Hi Phil, you wrote of the old photo of M13: “That old photo represented the cluster as it was seen more than half a century ago, a view we will literally never see again.”
It’s a nice clear image. If a picture had been taken of M13 every year since then from somewhere on Earth, and each image was scaled and placed in a film, say three frames per image, from “1930″ (or earlier) to now, it would be a short film, but would we see any subtle changes over that short stretch of time? (I ask because I was under the impression that globular cluster stars move around pretty fast in the interior.)
July 13th, 2007 at 1:26 am
Christopher Ambler Says: “So here’s my question - would you, given the couple million to spend, put it towards this, or some other project, if you had to choose?”
Interesting question, but there is more to it than just preserving old images from decay or loss due to dropping. Once the images are digitized, they can be digitally searched and analyzed. Think if Clyde Tombaugh had a digital analyzer rather than the manual “blink comparator” he probably would have found Pluto in a few minutes instead of years.
Back to the original post, I’ve logged enough hours in a darkroom over the years to appreciate you emulsion dilemma, but I don’t understand why the saliva from your tongue would be any better (or less worse) for the plate than the oils from your finger. Ditto your teeth marks.
- Jack
July 13th, 2007 at 2:02 am
Cameron,
I like the phrase, “quantum leapâ€. It implies discontinuity.
July 13th, 2007 at 4:23 am
Jack Hagerty says: “Back to the original post, I’ve logged enough hours in a darkroom over the years to appreciate you emulsion dilemma, but I don’t understand why the saliva from your tongue would be any better (or less worse) for the plate than the oils from your finger. Ditto your teeth marks.”
Because it’s sterile and because he likes the taste of it….
July 13th, 2007 at 6:17 am
I’m reminded of the book “When Worlds Collide” and the glass plates that have the proof the planets are coming are taken to New York via steamship.
July 13th, 2007 at 6:55 am
I sure hope they keep the originals, so they’ll still be able to read them in 20 years. Digitization will be helpful for analysis and presentation, but as a permanent archive? What format, and how long will we still have devices that can read that format? I have documents on 5-1/4″ floppies I created with my DOS-based word processor 20 years ago that I have a hard time reviewing today - unless I have hard copy in my files somewhere, which sorta defeats the purpose of electronic archiving.
(Anybody else remember the NPR “April Fools” story from a couple years ago, where they reported that the Library of Congress was transferring all modern CD music releases to 78s for permanent archiving?)
July 13th, 2007 at 7:22 am
DennyMo,
The challenge in Digital Archiving is two-fold: one : media will decay (almost always; see below). The first task is to keep moving the bits from one medium to the next , off Digital Videodisk onto DVDs, etc: I have old floppies that I can no longer read as they have bit errors, but I copied the contents to CD & hard driver long ago. Still have the bits, but .. two, document the formats, ensure archive formats (NOT Word .DOC, etc.) and have them as “self-documenting” as possible.
We have soft-copies of word-processing files, etc. from 20 years ago that we can no longer read - the format is unknown. We also have 4000 year old cuniform we can interpret…
I have a JPG of a block of cuniform buried somewhere at home. Its (apparently) from Mesopotamia, 4000 years ago, and a rant from a palace official (auditor of the time), complaining of this new-fangled papyrus and how it won’t last compared to good old reliable cuniform mud bricks.
The reason we have the cuniform today is that it was baked solid in a fire in the royal palace, that destroyed all the papyrus …
July 13th, 2007 at 9:17 am
|Think if Clyde Tombaugh had a digital analyzer rather than the manual “blink comparator†he probably would have found Pluto in a few minutes instead of years. |
It still takes a long time to take the pictures!
I remember reading in one of my first astronomy books that Tombaugh went back and found that an earlier sighting of Pluto had been masked by a defect in one the plates.
July 13th, 2007 at 9:36 am
I had heard some time ago about a gap in the Harvard College Observatory plate data created by director Donald Menzel, and when I read this entry I googled it and found this article:
http://www.nytimes.com/2007/07/10/science/10astro.html?pagewanted=all
“For all the hardships endured in collecting the data, the biggest threats these days have been bureaucratic. In the 1950s, a budget-wary observatory director, Donald Menzel, suspended the plate-taking operation for a while, resulting in what is now ruefully called “the Menzel gap.†And just a few years ago administrators, in search of more office space, ordered the collection compacted from four floors to three.”
The article mentions a number of other hardships the collection has faced over the years.
July 13th, 2007 at 9:39 am
D’oh!
July 13th, 2007 at 11:03 am
I take it the glass plates are not transparent? Otherwise it may not make a big difference which way around you put it in the telescope.
July 13th, 2007 at 11:32 am
The plates are transparent, but it’s still critical that the emulsion face the incoming light. If the light were to pass through the glass on its way to the emulsion, it’d be refracted and the image would be unnecessarily distorted. That, and that the glass itself will absorb some significant fraction of light, making the image darker, which is enough of a problem even without interference from the plate.
July 13th, 2007 at 1:50 pm
Also, the non-emulsion side had an anti-reflective coating on it to suppress internal reflections. So if that side faced the incoming light, you got even fewer photons. That happened to half my imaging class!
July 13th, 2007 at 2:03 pm
Digitizing the plates will make the data much more accessible but is there any movement to keep any portion of the plates as artifacts? The history of inquiry can be almost as fascinating, and culturally almost as important, as the results of that inquiry.
What does Harvard plan to do with half a million slides of glass?
July 13th, 2007 at 3:44 pm
Hi Phil
Digitizers are expensive. Perhaps the methods suggested bin the paper “Astrometry with Carte du Ciel plates, San Fernando zone. I. Digitization and measurement using a flatbed scanner.” (arXiv:0706.1900v1 [astro-ph] 13 Jun 2007) might make it possible for more people to contribute to capturing all the efforts of old astronomers.
July 13th, 2007 at 4:50 pm
sounds perfect for Google. They love scanning data and coming up with new ways to search the data. They work with universities to scan their libraries, why not their astronomical data?
July 13th, 2007 at 6:19 pm
If the cash can’t be raised for digitizing in situ, perhaps Plan B could be distributed storage & digitizing of the plates, something like “adopt a highway”. They could be offered to the public, with contractual agreements being signed that the recipients promise to make earnest effort at arranging for digitizing, at the individual’s expense, and to care for the plate at least until the data are in hand.
July 13th, 2007 at 7:01 pm
Beautiful!
July 13th, 2007 at 7:02 pm
Wonderful!
July 13th, 2007 at 7:24 pm
With the renovation of Griffith Observatory, two glass plates from the original Palomar Sky Survey are currently on display. They were delivered during an exhibit design meeting, and I remember the awed giddiness throughout the room as one was taken out to show us.
July 13th, 2007 at 7:44 pm
Phil said:
“A third one, weird as it may sound, is knowing which side had the emulsion on it! You can’t look — they’re light sensitive!”
I took a large number of plates (5×7) in the UVa parallax program during the ’70s. It was actually quite easy to determine the emulsion side of the plate. The way I did it, and what was taught to me when I started, was to moisten the tip of a finger and touch it to a corner of the plate. If the finger wanted to stick, that was the emulsion side.
Testing a very small area in the corner of the plate was safe because most of that area wouldn’t be exposed due to supporting the plate in the plate holder, and anything near the edge of the plate wouldn’t be measured anyway.
George
July 13th, 2007 at 8:34 pm
Yeah, used the inside of my upper lip even on cold nights at the Crossley telescope to test for the emulsion side. I’m of the very last generation to have used many plates, albeit only in grad school - at one time I could distinguish Kodak IIa from 103a emuslions by their taste (don’t try this at home, folks!) Sometimes the old plates have even more unexpected treasures. The double quasar 0957+561 showed up in a bunch of pictures from the start of the 20th century, at the edge of the field centered in the galaxy NGC 3079 - these images furnished the first evidence that the lensed quasar was variable and thus would be amenable to measuring a time delay (leading to an independent estimate of H0). I also used plates going back to 1895 in looking for evidence that quasars near bright galaxies might be gravitationally amplified things in the halos (answer: no, not unless halos are packed with black holes of 100 solar masses and up). Sobering to note that this bit of work is now 25 years old…
July 14th, 2007 at 9:10 am
You would think that after a hundred years of experience, the plate manufacturers would produce plates with a raised arrow or other indicator on the sides of the plate pointing to the emulsion side, that could be felt by the astronomer.
July 14th, 2007 at 10:43 am
A raised area on the surface of a plate would be a bad idea for the parallax plates I took. The emulsion side had to go flat against the yellow filter in the plate holder. (The 26 inch McCormick refractor was figured to be visual telescope since there was little to no photography being done in Astronomy when it was built.) And the back of the plate holder had to fit tightly. The when plate would be measured, emulsion side up, the back of the plate had to be flat on the stage of the measuring engine.
George
July 14th, 2007 at 10:58 pm
I meant on the side of the plate, not the front or the back. If a bump would cause problems, make it an indentation instead.
July 15th, 2007 at 1:50 am
Back in my grad school days, my advisor was visiting faculty at Berkeley, so we got weeks of time at the 120-inch at Lick. We used the Coude focus spectrometer’s 150-inch (as I recall) camera for stellar spectroscopy of bright OB stars in the blue to near UV. You wouldn’t believe the resolution, 0.13nm/mm (we looked at absorption of interstellar diatomic molecules against these nearly featureless stellar spectra and got rotational temperatures and upper limits to show that the 2.7K cosmic background emission discovered by Penzias and Wilson peaked in the right place to be black body. Had to digitize and add together over 30 plates from zeta Oph to get the final results. Did some other dimmer stars so not so many plates for them, but could show reasonable isotropy at least for CN). The plates we used (2 2×10 inchers–cut from 8×10s– end to end in the plateholder–which was curved for the whole length to be in the focal plane–had to be careful not break the glass) were sensitized IIa-O. My recollection is that only one person could sensitize properly, the oldest night asst, with the beard. I think it involved exposure to IR, then an extended bake-out. Relied on him to have the plates face up in the boxes. So all I had to do was load the plates into the holder, go through the light-lock doors, use my well-developed kinesthesis to negotiate the catwalk down to the camera focus, slide in the light safe slide and exchange holders. All of course in absolute darkness. BTW, in the plate loading room somebody had posted a sign ‘The East is Red’, the name of a popular song of the Cultural Revolution (this was in 1967-68). They posted it because it was literally true there.
Also James Lick is buried in the pier of the 36-inch Alvin Clarke refractor he paid for. Was still being used at that time for sky survey work, just like the Harvard plates. Another bit or two of Lick-lore: the 36-inch objective lens was being transported by mule up the mountain, slipped from its restraints, and rolled down hill and broke. Took two years to replace.
The mirror blank for the 120-inch reflector was a test blank for the 200-inch at Palomar (donated by Corning Glass). The whole telescope, including spectrograph and dome cost about $2.5M by the time it was completed in the beginning of the 50’s. Th blank was thin-ish, hence the long-ish focal length of the telescope–about f/5 or f/6. One night we had Werner von Braun with us to get an idea of telescopes at the time they first were conceiving of what became the Hubble. Which cost a wee bit more than the 120-inch, as I recall. We were fogged in all night and couldn’t even open the dome. The nice ladies in the kitchen were afraid NASA was going to shoot their telescope into orbit. He was very polite and assured them it was not so.
There was a packrat, but organized, chief astronomer there some decades before my visits who had a wall of drawers for oddments. Supposedly there was one drawer marked ,’String too short to use’.
A suggestion, if you haven’t already done it, might be to have an occasional post on telescope history and personal reminiscences from student days and early careers–the olden times of glass plates and such. This comment might provide a start of one such post.
July 15th, 2007 at 2:09 am
Oh, and the kicker in my work, talking about perishibility of digital stuff, is that somebody forgot to put in the write-protect ring on the tape of the final syntheses of the spectra. This was the ultimate result of more than a year’s effort on my part using the huge FORTRAN program I wrote to crunch the numbers, converting densities to intensity and calibrating wavelength from the iron spectrum recorded on the plate, then correcting for proper motion and adding.
The tape got written over a couple of years later and only the hardcopy graphical printouts were left. The plates were sent back to the Lick archive. Maybe I’ll visit them someday.
Another by-the-way. A lot of raw data tapes were erased in a NASA archive just before my time. Turned out the motor from a big floor buffer had a big enough AC magnetic field to ruin almost every tape on the bottom shelf.
July 16th, 2007 at 8:19 pm
Scott said:
“I meant on the side of the plate”
Sorry I missed that interpretation. But the plates I used were not very thick. Let’s see, it is 4:00 AM some January morning and I need to load the plate holder. I take off my gloves
and run my finger along a thin edge looking for an indentation in the shape of a arrow. Why
don’t I just moisten my finger tip and tocuh a corner on one side? I’ll indentify the emulsion, get the plate loaded and my gloves back on, before I could find your arrow I bet.
The real point is though, as a couple of others sort of mentioned, it was really very easy to determine the emulsion side of the plate, if you did it more than two or three times, not
just once or twice for a class as some sort of “Rite of Passage”. The real point is why complicate the manufacturing process of the plates to make it slightly easier, at best, to do something which was already easy?
George
July 17th, 2007 at 3:50 am
VJB - the ring on open reel 1/2 tape is a write enable ring, not a write protect ring, so someone must have forgotten to remove it. (Or put one back in.) Used to have hundreds of them, but a few years ago I went looking for one for a friend who wanted one to use to construct a geeky christmas tree ornament and it took hours to find one! They are also a good test of antique computer knowledge. Most people seem to recognize an IBM punch card (though they might not recognize the term “Hollerith code”), but write rings are more obscure. I wonder if the 029 key punches are still in the room down the hall, up the stairs, and around the corner from the plate stacks?
July 18th, 2007 at 2:49 pm
Buzz Parsec–you are undoubtedly right about the rings. At our NASA facility at Columbia (where Jim Hanson came on as a post-doc with hair while I was there) only the High Priests of IT ever actually even got to touch tapes (except for the 7-track ones we used on the scanning microdensitometer built for my project and probably never used since). Incidentally, the company who built the digital interface for the recording system for this instrument was a little firm out on Long Island called Computer Associates (yes, THAT one) in its early days. All the logic was DTL in the Fairchild epoxy-sealed TO-5 like packages that were the predecessor to DIPs. All hand wired; no wirewrap.
July 18th, 2007 at 3:07 pm
Oh, and while I on the trip down memory lane, NASA’s Goddard Institute for Space Studies, which I refer to, still had a 7094 vacuum tube computer when it was still located in the Interchurch Center when I started there. When we moved to 2880 Bway (over the now-famous Tom’s Restaurant) we got an almost super-computer, an IBM 360/195. It had I recall 4 megawords of semiconductor memory (4 9-bit bytes including parity). It needed water cooling. It was to come online for general use after Labor Day 1966 (I think that’s the year), but over the holiday weekend, the cooling sprung a leak. Trouble ensued, and it never quite was up to snuff. This machine had the old ‘washing machine’ winchester hard disk drives with the removable platter stacks. The best thing, though were the IBM 2250 graphical display consoles (rental to IBM was $5K or $15K a month). I did all my plate calibrations with these, as I had to pick the iron lines for spectrum wavelength calibration, etc. I got into interactive programming much earlier than most people as a consequence, as there seemed to be no reasonable way to recognize line patterns. At least with in the limitations on my gigantic FORTRAN program (3600 cards–almost two whole boxes–WOW!) Now I sneer. The main applications program for the instrumentation my little company builds is now up to half a million lines.