The The Cyclopædia; or, Universal Dictionary of Arts, Sciences, and Literature edited by Abraham Rees (London: Longman, Hurst, Rees, Orme, & Brown, 1819-1820) is one of the great historical and technological resources of the first industrial revolution. In its original British edition it runs to 39 volumes of text and six additional volumes of Plates.
For a general introduction to this work, see the English-language Wikipedia article on it at: https://en.wikipedia.org/wiki/Rees%27s_Cyclop%C3%A6dia . (But note that there are problems with the list of links to digital volumes at the end of this Wikipedia article.)
Rees' Cyclopædia can be difficult to use. It is unpaginated. It has no overall index (even though aspects of some subjects, such as technical horology, are divided among multiple articles). You have to know the sometimes idiosyncratic name for your topic before you can find it. (Typefounding, for instance, is "Foundery, Letter".) There is a catalog of the plates and an index to the plates (both in Vol. 39), which can help. But the plates themselves are gathered into topical sections in the Plates volumes and are not closely associated with their articles. (There are 50 plates on Horology, for example, which illustrate a dozen or more different articles spread throughout the 39 text volumes. Other plates are gathered in a "Miscellany" - the plate of the typefounder's hand mold is one example.) Moreover, the numbers on the plates contain errors and duplications.
Original printed copies of Rees are now expensive antiquarian items not available to the average independent researcher. But these issues with the printed version are all present in the digital facsimiles as well.
However, there is no complete run of digitizations of all 39 volumes of text and 6 volumes of plates from any of these three good sources individually. There isn't even a complete run as a composite of these three (volumes 11 and 15 are missing from all three). Most of the volumes from these three good sources are hosted on The Internet Archive ("IA") - although finding them there can be difficult because there are no single-page indexes of the IA volumes. A composite full run (which still misses many items, including most of the UC volumes) is indexed at the Biodiversity Heritage Library ("BHL"). But one of the three volumes actually digitized by the MBG is available only at the BHL, not the IA. There are also several poor Google Books ("GB") digitizations on both the IA and the BHL which are mistakenly cataloged as MBG digitizations.
Then there are problems with the plates. For example, Vol. 2 of the plates is available in only one good digitization (UT, via the IA or the BHL). But the physical copy at UT from which their scan was made is missing two of the plates in the Horology section (a section of particular interest to me, as it happens). At present, these can only be seein in one of the poor-quality GB digitizations.
These GB digitizations are adequate for the text volumes, but not for the plates. Fortunately none of the plates for Rees fold out (Google's digitization project is intended to capture texts only, so they never fold out the plates.) But the images have been so heavily processed (sometimes to bi-level) that the plates are of exceedingly poor quality, and many of the figures on them are completely useless. But for volumes 11 and 15, for which there are no good digitizations, the GB versions are all we have.
The versions available at the Hathi Trust ("HT") constitute a composite run. Some of them (identified as those from the University of California) are actually the IA scans of the UC copies. Others (identified as those from the University of Michigan) are GB scans. As noted earlier, some of these GB scans of UM copies have been uploaded to the IA and cataloged as if they were MBG scans. This can be quite confusing (but look for the big Hathi elephant logo on the first PDF page of the volume).
There was also an American edition of Rees (see the Wikipedia article on the Cyclopædia for more information on it). This has been digitized by Google from the Princeton University ("PU") copy. It is available from the HT.
Finding the volume you want can be most confusing, especially if you care about sources. The BHL at least has a single page which lists all of the volumes. Its listing, though, is of a composite run of digitizations from UT, UC, MBG, and HT GB digitizations of UM copies miscataloged as MBG digitizations. The IA has a similar composite run, but no overall index page. To find the volumes on the IA, you must search for them (search using both "Cyclopaedia" and "Cyclopædia"; "Cyclopedia" won't work).
The Wikipedia article on the Cyclopædia is in general good, but its list of links to the British edition on the IA is out of date: it fails to link to some UT digitizations which really are there, links to one UT digitization which isn't there (though this may be a temporary metadata problem at the IA), misses most of the UC digitizations, and doesn't distinguish UT digitizations from MBG ones. There are also some minor errors in it citation of first and last articles for a few of the volumes.
The image below presents three digital versions of the same engraved figure. It's Fig. 3 from Plates volume II, Horology Plate XIII. The left image is by the University of Toronto. The middle image is by Google of the Österreichische Nationalbibliothek (Austrian National Library) copy, as presented by the ONB themselves. The right image is also by Google, is from the same digitization of the same ONB volume, but it is as presented by Google Books.
The UT version is perfectly legible: you can see all of the lines of the original engraving and read all of the characters. The Google/ONB version is marginally useful. You can make out all of the components and features of the device, and read most (but not all) of the characters. The "pure Google" version is useless; you can't even discern the basic features of the device.
I have probably missed some volumes. Indeed, for the good-quality scans I hope that I've missed some, since no run is complete. In particular, we need second sources of good quality for all of the Plates volumes (except maybe Plates Vol. 3, which is already represented from UT and MBG digitizations).
1. The Internet Archive ( archive.org)
The IA has the deepest collection of volumes of Rees. The particular advantage of the IA is that if original scan page images are available, they are presented. To find them, go to the "Download Options" panel and click on the "Show All" button. You'll go to a list of available files. In general, the biggest file is the original version and all of the other files are derived versions (or metadata about the files). So, for example, for a digitization done by the University of Toronto, the file ending in "_orig_jp2.tar" contains the page images from the camera (converted lossily into JP2 format, but a first-level onversion such as this doesn't hurt the image quality). If you unpack this file, you can see parts of the photographic rig in the background. The file ending in "_jp2.zip" contains processed versions of these images (and a second level of lossy processing, giving lower quality images). The file ending in ".pdf" is a PDF encapsulation of further (lossy) conversion. It's quick and easy to read, but the image quality is visible degraded. (The ".djvu" file is a DjVu format conversion - this should never be used. See the note on Undetectable Data Corruption in JB2/JBIG2; DjVu uses JB2).
This has a subset of the same collection of digitizations that the IA has. Here's the overall BHL page for Rees, including volumes sourced from UT, UC, MBG, and HT GB digitizations of UM copies mistakenly labeled as being from the MBG. Its overall index is nice: http://www.biodiversitylibrary.org/bibliography/59683#/summary
The six volumes listed (some in truth, some in error) by both the IA and the BHL as being from the MBG (text: 11, 15, 17, 25; plates: 3, 6) are also listed in the MBG's own digital library, Botanicus. At the present time (2015), however, they won't actually display, due apparently to technical issues. See: http://www.botanicus.org/bibliography/b11689535
They have the digitization done by Google of their copy. However, as can be seen in the image comparison above, their presentation of this digitization is of considerably higher quality than Google's. Unfortunately, they license their version only for personal noncommercial use (which means that I can't use them here to supply plates missing from the UT digitization). Their "Digital Reading Room" is at: http://www.onb.ac.at/ev/digital_readingroom.htm You can find Rees by searching for rees cyclopaedia (with 'ae' as two characters).
The Hathi Trust has catalog pages for both the British and American editions. The British edition as hosted by the HT consists of the (incomplete) UC volumes (the same ones which are at the IA, I believe) and the GB UM digitizations (with overlap in coverage between the two). Image download is limited to individual pages (and rate-limited within that) unless you are with a participating institution. Their catalog page for the British edition is at: http://catalog.hathitrust.org/Record/001464694
This isn't a repository itself, but the article on the Cyclopædia has a list of links to entries for both the British and American editions on the IA (British) and the HT (American). But, as noted earlier, the list of links into the British edition has some issues. It can be useful at first, but you're as likely to miss things as find them (e.g., the UC digitizations, or a non-Hathi version of the Atlas). The page is at: https://en.wikipedia.org/wiki/Rees%27s_Cyclop%C3%A6dia . If you can't get to the Wikipedia page on Rees' Cyclopædia using the link above (it contains non-ASCII characters and may become corrupted in transit), then search the English-language Wikipedia for "Rees" to find the page on Abraham Rees. In it, click on the link to the Cyclopædia.
These are the volumes that I have located on the Internet Archive which were digitized from the University of Toronto's Robarts Library. These digitizations are of reasonably high quality (though for the Plates you probably want to dig down and use the original JP2 page images). This is the most complete good quality digitization. It's missing only Volumes 11, 15, 17, & 25 of the texts and Volume 6 of the plates. This is the best nearly-complete run; start here.
These are all of the volumes of Rees (of which I am aware) actually digitized by the MBG. Text Vol. 17 and Plate Vol. 6 are available at both the IA (see above) and the BHL. Plate Vol 3 is available only at the BHL (as far as I can tell).
These volumes are cataloged on both the IA and the BHL as if they were MBG digitizations, but they are not. They are GB digitizations of UM copies, processed by the HT. As they are GB digitizations, their image quality is quite poor.
The online digital library of the Missouri Botanical Garden, "Botanicus," lists four volumes of Rees texts scanned (Vols 11, 15, 17, and 25) and two volumes of plates (Plate Vols 3 and 6). These are the same volumes associated with the MBG at the IA and BHL (partly correctly, and partly in error, as noted above). Unfortunately, at least at the present time none of these volumes will load at Botanicus (tested in various browsers under Linux and Windows). (A look at the page source reveals that they're using third-party software to serve ways of viewing remote images, rather than simply serving the images. Complex solutions tend to fail.)
For the Botanicus entry for these, see: http://www.botanicus.org/bibliography/b11689535
Here are all of the volumes that I can find on the Internet Archive digitized from University of California sources. These are good scans. It's a shame that this series isn't more complete. (As far as I can discover, they contain none of the four text volumes missing from the UT scans (11, 15, 17, 25) and none of the Plate volumes at all.)
As noted earlier, the Österreichische Nationalbibliothek (Austrian National Library) has the digitization done by Google of their copy. As presented, it is of higher quality than the Google presentation of the same digitization. However, their license terms permit only personal, noncommercial use (so I cannot reprint them here for convenience). Their "Digital Reading Room" is at: http://www.onb.ac.at/ev/digital_readingroom.htm You can find Rees by searching for rees cyclopaedia (with 'ae' as two characters).
The easiest (and probably the best) way for the modern practical horologist to access the horological material of Rees' Cyclopædia is to acquire a physical copy of the volume of extracts published in 1970 under the title Rees' Clocks, Watches and Chronometers (London: David & Charles (Publishers) Ltd., 1970; Rutlant, VT: C. E. Tuttle Co., 1970). This book gathers together in a single location almost all (it omits astronomical machinery) of the articles related to horology. It splits up the section of 50 plates and distributes the plates so that they accompany their relevant articles. It also contains a (new) index to these texts.
While secondhand copies of the David & Charles collection are still surprisingly cheap (in 2015), this will not always be so. I'd also like to have a digital version in any case, just to supplement my physical copy. So here I'll assemble what is effectively a digital edition of the David & Charles reprint, using the currently-best-available digital sources. It won't be entirely the same, but it will be similar. (I can't reprint the 1970 index of the David & Charles book, but I can add at articles omitted from it, such as "Planetary Mechanics.")
Before we can assemble an edition of "Rees for Horologists," we have to confront the issues presented by its plates. There are 50 of them (plates, not issues). In the original printed edition these were gathered into a single section, entitled "Horology," in Volume 2 of the volumes of Plates. These original plates contain an error in numbering. They run from I (Ancient Clepsydrae) through XXXI (Dial Work), XXXII (New Dial Work), and XXXIII (Dial Work, with a fish). Then at the 34th plate the numbering jumps back to XXXI (Escapements) and continues on to the end, Plate XLVII (Repeating Watches), which is really the 50th plate.
The digital versions introduce further complications. First, the only good quality digitization is that of the University of Toronto ("UT"). But either the physical copy at UT is missing two plates (Plate III (Clock-Movement) and Plate XXIX (Compensation Balances)) or they missed them in digitizing the volume. They're absent from the digital version.
This means that we have to supply them from one of the three Google Books ("GB") digitizations (of copies from Stanford University ("SU"), the Österreichische Nationalbibliothek (Austrian National Library) ("ONB") as presented by Google, or the University of Michigan ("UM")). (I cannot use the ONB copies as they present them directly on their own library's website, as their terms of licensing permit personal use only but what I'm doing here is reprinting.) These GB digitizations are, regrettably, of much lower quality. These digital editions are also hard to navigate, because in all three there are duplicated and interpolated plates (but none missing, so far as I can tell).
In the image comparison below, the left image is taken from the "_orig_jp2.tar" images. Already there has been loss in the image compression, but we can't do anything about this. (What's really happened by this point is that sensors in a digital camera - arranged in ways that really don't look like a bitmap at all - together with the software in the camera have produced, through a fair bit of image processing, a bitmap. This is output from the camera in some format (possibly a lossless format, but more likely a lossy format such as JPEG). If the camera's output format was not already JP2, then this image has been further re-saved, lossily, to JP2. So already by this "original" image there are many layers of digital image processing and lossy image compression.)
The image in the middle was created (by either UT or the IA) from the left "original" image (presented in the "_jp2.zip" file). It's been rendered from JP2 to an internal bitmap format, manipulated (rotated, at least), and then re-saved into another JP2 image. Every time you save or re-save to a lossy format (JPEG, JP2, etc.) you aren't really saving the image - you can never get your original bits back. Rather, you're creating an entirely new image which just happens (by an amazing bit of mathematics) to look almost exactly like your original image. In less technical terms, each time you save lossily, you introduce image degredation. Doing this once, or even twice, is usually ok.
By the third (rightmost) image, though, the image degredation is not only visible, but annoying. In this third image, the middle image (which is at least a second-generation JP2) has been re-saved yet again as another JP2 (JPEG2000, aka JPX) and embedded within a PDF document.
Here are the plates on Horology, as individual images. I've assigned each of them a "sequence number" as they appeared (or should have appeared) in the original printed edition. All of these except Seq. 3 and Seq 29 are from the UT digitization.
Click on any of the images below for a larger version. For UT sourced images, the left image links to a PNG rendering of the "original" JP2 image, rotated 90 degrees (lossles rotation, after rendering to PNG) as appropriate. They aren't cleaned up in any other way (they're not rotated so as to be exactly vertical, nor are they cropped to eliminate the background of the photographic rig). For UT sourced images, the right image links to a PNG rendering of the IA/UT processed JP2 image. In each case, text links to the JP2 versions appear below (at present (2015) many browsers still can't support JP2/JPEG2000/JPX, so these may not display correctly if you click on them).
Not present in UT digitization. The image here is from the UM digitization. In this instance, the image below links to a PDF of the single page, extracted from the GB/SU PDF. (Aside: In this instance, the ONB-via-ONB image isn't quite as good as the UM digitization.)
(Here's a link to the same, rendered at an unnecessarily high 600dpi, as an image file: rees-cyclopaedia-plates-2-google--XxMAAAAMAAJ-mich-img384-plate-seq-03-clock-movement-render600dpi.png )
Not present in UT digitization. The image here is from the SU digitization. It's the least-bad of the three freely redistributable versions (UM, SU, ONB via GB), but it's still pretty bad. The ONB-via-ONB version (not freely redistributable) is slightly better, but not good. In this instance, the image below links to a PDF of the single page, extracted from the GB/SU PDF.
(Here's a link to the same, rendered at an unnecessarily high 600dpi, as an image file: rees-cyclopaedia-plates-2-google-pNo1AQAAMAAJ-stanford-img450-plate-seq-29-compensation-balances-render600dpi.png )
Rees for Horologists: The Plates
This is a composite digital edition. It consists of the processed JP2 images of the UT/IA digitization (rendered to PNG to forestall further loss), together with Google Books digitizations for the two plates missing from the UT copy.
It is a 437 Megabyte PDF file.
This article appears on pp. 3-9 of the David & Charles book, with Plates 5/V, 6/VI, and 7/VII. (I'll cite both sequence number and the number appearing on each plate. Here it doesn't matter; later it will.) The original article appeared in Text Vol. 7. It's on images 650 - 653 of the IA/UT PDF.
Rees for Horologists: Chimes
This article was omitted from the David & Charles book. It appears immediately after the main article "Chronometer" (see above), under the title "Chronometer. - Exemplification of its use in determining the longitude of a ship or place." I don't believe that it references any plates (though I confess I have not read it in full detail). It does include several tables and a series of solved problems. It occupies images 45-66 of the IA/UT PDF.
"Foundery, Letter, or the Method of Casting Printing Letters." Text Vol. 15. PDF sequence numbers 125-128 in the GB scan of the UM copy which is miscatalogued at the IA as MBG (https://archive.org/details/mobot31753002000526)
All material from Rees' Cyclopædia reprinted here is in the public domain. The reprints here remain in the public domain.
All portions of this document not noted otherwise are Copyright © 2015 by David M. MacMillan and Rollande Krandall.
Circuitous Root is a Registered Trademark of David M. MacMillan and Rollande Krandall.
This work is licensed under the Creative Commons "Attribution - ShareAlike" license. See http://creativecommons.org/licenses/by-sa/3.0/ for its terms.
Presented originally by Circuitous Root®