Adventures in the HCO Logbooks

2024 September 19

The DASCH scanning project may be officially complete, but that doesn’t mean that every single plate held by the Harvard College Observatory (HCO) Plate Stacks has been digitized. As part of preparations to squeeze in a few more scans of some “odds and ends” plates, I’ve been digging into the Plate Stacks’ collection of historical observing logbooks to learn more about them. The logbooks provide a fascinating window into how cutting-edge science was performed, a hundred and twenty years ago.

It’s absolutely true that the point of DASCH was to digitize the Harvard plates to support time-domain astronomy projects. But, as often happens when you take on a big project like this, you find yourself heading in surprising directions. You can’t do time-domain science without knowing when each plate was made, and that information isn’t recorded on the plates themselves. (How could it be?) Timestamps, and lots of other really useful metadata, were recorded in handwritten observing logbooks that accompany the plates. These logbooks end up being hardly any less important to doing science with the Plate Stacks than the plates themselves.

And so a major part of the DASCH project was an effort to digitize these logbooks. This started in 2006 and was led by volunteer George Champine, who built a photo station and photographed all of the books himself. How many? Try 813 physical volumes, totaling around 185,000 pages of material — a true tour de force.

In partnership with Harvard’s Library Imaging Services, the DASCH team obtained improved imaging of some of these logbooks as well as new digitizations of the personal research notebooks of Annie Jump Cannon and Henrietta Swan Leavitt. (These notebooks don't record individual observations, but they do capture the work of two legendary astronomers, and often reference specific plates in the Harvard collection.) The final photography collection covers 842 volumes, 166,062 image files, and 1.8 terabytes of data.

All of these images are made available as part of the DASCH data products, although they’re not convenient to browse at the moment. For instance, this directory listing gives an index of all of the photographs of the 15th volume of the “D” series of plates, which are somewhat unusual because they’re photographic duplicates of other plates, rather than original exposures.

I’ve spent a lot of time in the D logbooks lately because many of our odds-and-ends plates come from the duplicates series. This is understandable — precisely because they’re duplicates of other plates, not many of these will necessarily improve DASCH’s utility for time-domain astronomy. But it’s still valuable to create digital surrogates of these objects.

What I quickly noticed was that they seemed to be duplicating the same plates over and over again. Notice all of the ditto marks in the “Orig. Pl. No.” column of this page:

Based on our inventory, our duplicates cabinet appears to include no fewer than 21 copies of plate A3393! Seems a little excessive, no?

What I realized, halfway through my trawl of the D notebooks, was that these log entries give us a detailed look into the social history of astronomy. Everyone wanted a copy of A3393! Which turns out to make sense — it was one of the plates used for Leavitt’s research into SMC Cepheids, leading to the Leavitt Law and the understanding that the universe was much bigger than the Milky Way. Really, these duplicate logbooks are something like the access logs for a present-day webserver — individual transactions that collectively tell us where the zeitgeist is at. Of course there was a huge demand for plates of Halley’s Comet right after its passage in April/May 1910:

You can also see hints of a history that’s much smaller, in a certain sense. Was the person working in the darkroom on New Year’s Eve, 1927 annoyed that they had to be there making a copy of MC22997? Did they go to a good party after the day was done?

But as far as DASCH was concerned, the point of photographing the logbooks was to be able to extract observing metadata about all of the plates. So as the logbooks were photographed, their text was transcribed into data tables. This was its own major undertaking, involving contributions from volunteers local to the CfA, at AMNH, the Astronomical League, and the Smithsonian Transcription Center (STC), as well as commercial services. These “logindex” data tables aren’t online, but their contents form the basis of the StarGlass database as well as virtually all DASCH data products relating to the plates. Beyond DASCH’s focus area, work is ongoing at the STC to transcribe the research notebooks of the early HCO astronomers, particularly the Women Astronomical Computers, as part of Project PHaEDRA.

Some of my recent projects, however, have hewed close to the traditional DASCH activities. Our odds-and-ends collection includes a few dozen plates from a short-lived “Y” series, obtained in Cambridge using the 8-inch Clark Doublet telescope in 1888. It appears unlikely that these plates will have much scientific value, but as long as we have them, I’d like to digitize them and assemble the highest-quality metadata that I can. This raises its own challenges.

For instance, as I was entering the Y-series data, I thought that I had everything that I needed, but was very concerned by this entry for plate Y29:

Dated June 2, 1888, this seems to say that the exposure for Y29 started at “7h23m (?)”. The written annotation says something like, “Taken 3m after five [?] stars are visible ... Polaris seen 7 41.” The question is, what kind of time is that “7h23m”? From the prior entries, I had thought that such timestamps were giving local sidereal time, but on that day an LST around 7 hours would correspond to around 2 PM — not exactly nighttime! Likewise, it’s clearly not a local time — stars would only start coming out around 9 or 10 PM on a June night in Cambridge, so even if the time had an implicit “PM” attached, it wouldn’t work.

After a great deal of flailing around, I was able to get some traction thanks to a remark in “Investigations in Astronomical Photography”, an 1895 volume from the Annals of Harvard College Observatory. I was only able to find this remark thanks to an entirely separate imaging and transcription effort — ADS’s unrelenting work to organize and index the pre-digital written history of scientific astronomy. Currently, ADS appears to have digitized 626 publications in the Annals — some of which are hundreds of pages long individually — as well as 515 articles in the Harvard College Observatory Circular. I’ve found that these publication series provide invaluable contextual information to help understand the Harvard plate collection.

In my particular case, descriptions of some lunar photographs give precise timestamps of certain plates made and logged contemporaneously with the Y series:

Various cross-checks convince me that this description refers to plate X00981 or X00982, logged here:

Other examples give consistent results: if the logbook timestamp system is TSS, we have TSS = GMT - 8.

But wait! 16:00 GMT is 11 AM Eastern Time — once again, not a great time of day for astronomical photography. Now what’s going on? This one is a bit easier to solve: before the introduction of the Universal Time in 1928, GMT days could interchangeably start at either noon or midnight, depending situationally. (Thanks, Wikipedia!) So the Pickerings’ GMT must start at noon. In that case, the logbooks’ timestamp system must be measuring ... hours after 3 PM Eastern Time! I don’t understand this choice at all, but it’s consistent with the evidence: it would put the Y29 observing night as starting around 10:20 PM, not unreasonable for a Cambridge night near the summer solstice.

In case you’re wondering, the now-standard US timezone system and GMT as the world standard were adopted in 1883 and 1884, so the Y-series observations were happening just a few years after these conventions were codified. I can imagine that various other aspects of timekeeping that we now take for granted were still in flux, although it’s still hard for me to see why the Cambridge times would be referenced to 3 PM local time. It occurred to me that the timing system is equivalent to counting hours after noon in the Pacific zone, which is “more round” in a certain sense, but I have a great deal of trouble seeing how it would make sense to anchor your Cambridge time standard to the West Coast. At least one Cambridge telescope was shipped to California by the following year (1889), but it would seem utterly bizarre to adopt a Pacific-oriented timing in the logbooks in advance of the shipment. Anyway, it’s a bit silly for me to get too deep into speculation — I’m sure that there are experts in early-modern timekeeping who could shed light on all this.

Of course, it then turns out that halfway through the X02b logbook, they switch to timestamping using LSTs. Then, some of those are off by 12 hours, a problem that apparently was observed in a variety of the Harvard logbooks.

While I’m undertaking my investigations for fairly narrow purposes, they underscore to me just what level of historical richness is centered on the Plate Stacks. The DASCH dataset represents just one piece of it, and depending on your perspective it certainly needn’t be considered the most important piece. There are the glass plates; their jackets; the observing logbooks; the astronomers’ notebooks; card catalogues and looseleaf logs that I haven’t even mentioned yet; the publications in the Annals and other venues; and a whole wealth of other historical material spidering out from there. Hundreds of thousands of photographs and pages of written material. Hundreds of human lives. A whole Glass Universe, if you will. The plate stacks are a wonderful historical resource, representing a unique intersection of science, technology, and society, and we’ve barely scratched the surface of what they have to offer.