Diderot & d'Alembert

Finding the Encyclopédie

image link-topic-sf0.jpg

... it shouldn't be this hard!


1. The Availability of the Encyclopédie

Diderot, Denis and Jean le Rond d'Alembert, eds. Encyclopédie, ou Dictionnaire Raisonné des Sciences, des Arts et des Métiers, par une Societé de Gens de Lettres. (Paris and [false imprint at] Neufchastel [Neuchâtel]: [various publishers], 1751-1772)

The initial publication history of Diderot and d'Alembert's Encyclopédie was made complex not only because of the issues involved with its massive scale but because of conservative opposition to such a liberal and subversive work. The basic text was published in seventeen volumes from 1751 to 1765, together with eleven volumes of plates ("planches" in French) from 1762 to 1772. (The volumes of plates are numbered one through ten, but the second volume of plates appeared in two parts.) The first seven volumes of the text were published in Paris from 1751 to 1757, but permission to publish in France was then withdrawn. The last ten volumes of the text were published with a false imprint of "Neufchatel" (modern: Neuchâtel) in 1765. The volumes of plates all appeared with a Paris imprint from 1762 to 1772. There were other related volumes as well (the prospectus, errata, supplmenents), and later additions which may or may not be considered canonical. Some digital reprints (such as that of the ARTFL project) number the regular and plate volumes sequentially (in such a scheme the plates begin with "volume 18").

As important as this work is, prior to the digital era access to it has been difficult due to its scale. An original copy of the first edition is now an expensive antiquarian item (I did a quick search online, and found auction records of one copy that sold for $94,000 back in 1998.) There were reprints in the 18th century (e.g., in Geneva), but these are now as inaccessible as the original. A selection of the plates (only; no text) was published in 1959 by Dover Publications, in two volumes, with restrictions on their use. Moreover, in this reprint the plates on any given subject were often incomplete (and carried no indication of what was missing). There was a microfilm version produced in 1969, with a printed edition (done 4-up to a page), but copies of this now cost several hundred dollars.

Access to digital versions of most of the original edition, both text and plates, is now possible. (The final volume of plates (10th issue, 11th volume of plates overall) does not seem to have been digitized anywhere, however.) But it in all cases there are issues which complicate, unnecessarily, simple scholarly access to one of the most important books in the history of the western world.

Here I'll summarize my current knowledge of how to access this book online, starting with the most useful sources.

2. Via the Internet Archive (Ottawa scans)

[Good full text, poor plates]

3. Via ARTFL

[Very good transcribed edition. Good linked page images (text and plates), but title pages missing from Vol. 9 on. No presentation of the text as volumes.]

The ARTFL Encyclopédie Project of the University of Chicago is in many ways the best, and most accessible, version of the Encyclopédie. Yet, though I do not wish to appear ungrateful for it, there are several issues with it which must be recognized.

The overall format of the project is that of a database with a web-based interface. The database contains a transcription of the Encyclopédie together with images of the original pages. You get to the page images by searching (or browsing) for something, and then clicking on a link in the transcribed text to see the original page. This is all pretty well done, and it is an extremely useful way to search the work.

They do not, however, provide any version which gathers these page scans into entire volumes. This can produce a sense of disorientation, as the work is cut loose from its original format as a book.

They do provide a fine discussion of the provenance and the issues with their source for the text (and their careful resolution of these issues). See:

(Lexicographers might note that the first of these contains a startling use of "expertise" as a verb.)

For the source for their transcribed text, they used the 1969 microfilm version (Leiden, The Netherlands: IDC, 1969). This is the same version that was produced in print as a "compact edition" (NY: Pergamon Press, 1969). They've gone to considerable lengths to verify that this IDC/Pergamon edition is in fact a reliable version of the original edition.

It is puzzling, therefore, that they never actually give the source for their page images. One can presume that it is also the IDC/Pergamon edition, but they never actually say so.

The other oddity in the ARTFL edition (one that it shares with the imagery on Wikimedia Commons) is that volumes 9 through 17 of the text lack all front matter. The title page for Vol. 8 serves as the title page for all of them (even though it is clearly just the title page for "Tome Huitieme, H-IT"). The digital versions presented by the "Gallica" site of the Bibliothèque nationale de France (see below) confirm that volumes 9-17 did have their own title pages.

(A comparison with the IDL/Pergamon edition (which I don't have) would be interesting; if they, too, lacked these title pages then that would be a good indication that the IDL/Pergamon edition was the source for these page images.)

The ARTFL site itself (that is, the transcriptions and the overall mechanics of the site) are in copyright, but that isn't a serious issue. ARTFL is silent on the matter of the copyright status of the page images that they supply; they do not claim any new rights. The University of Michigan Collaborative Translation Project (see below) is of the opinion that these JPEG page scans are in the public domain.

4. The UM Collaborative Translation Project

[In-process (it's a big task!) Usability and longevity issues due to lack of open license.]

There is a collaborative online translation (into English) project underway, hosted at the University of Michigan:

This is a wonderful thing, of course, but it has two issues. First, it is not yet complete. Second, and more importantly, the rights to each translated portion remain with the translator; there is no overall free licensing of the project. While this certainly respects the rights of each individual translator, and while in theory it would be possible to contact the translator to request permission to use their work, in practice it locks this collaborative effort into its online presentation. (This also guarantees its demise at such time as the University of Michigan library budget finds it redundant. All texts perish, to be sure, but university budgets perish sooner.)

Still, while it exists, this collaborative translation is very useful. It does link back to the original scans, as well (from the ARTFL project).

5. Via Gallica / BnF

[Incomplete; reasonable scans; hard to find particular volumes.]

The natural source for the Encyclopédie is, of course, the Bibliothèque nationale de France. However, they do not seem to have all of it online via their "Gallica" digital library website. (Volumes 6, 7, 8, 10, and 11 are missing, as is the last volume of plates.) The folks at the French Language Wikisource's page on the Encyclopédie can't seem to find them at Gallica, either.)

The volumes which are present have several advantages, though:

They may be read online or downloaded in PDF format.

The scans are obviously different from those presented at the ARTFL project (it's nice to have multiple sources). The ARTFL scans are cleaner, though.

An examination of a sample page from Vol. 1 of the text of this version indicates that the PDF filter "CCITTFaxDecode" was used for a single page image. This, in turn, indicates that either CCITT Group 3 or Group 4 compression was used. This is good, because it avoids the risk of loss of original information posed by the use of JBIG2 or JB2 compression in Google Books PDFs and in DjVu-format documents. In this sense, these digital versions are archival (which is what one would expect from a national library).

They do assert ownership of the digital version, but permit noncommercial use freely.

The only real problem with the Gallica/BnF version is that it is difficult to find particular volumes in their catalog. They use the full title of the Encyclopédie, which is long, which is very similar to the titles of several other encyclopedias, and which takes up most of the space in the display of a catalog record. They do not include the individual volume numbers in the catalog records. You have to read (online, at least) the actual title page of each volume to find out which one you're looking at.

Because these are so difficult to locate in the Gallica/BnF catalog, here are my local copies of versions that I've downloaded. PLEASE NOTE that these are restricted by license, and that only their noncommercial use is free. These digital versions are owned by the French state, a nuclear power. My use of them here is entirely noncommercial.

Note also that I have been unable to locate Vols 6, 7, 8, 10, and 11 of the text and Vol. 10 (11th volume) of the plates on Gallica.

6. Via Wikimedia Commons and the French Wikisource

[Complete except for last plate volume. Encoding issues.]

Wikimedia Commons contains the seventeen volumes of the text of the original edition and ten of the eleven volumes of the plates:


The texts are presented as sequences of JPG images, and the plates as sequences of PNG images. But the JPG images for the text for volumes 9 onward do not include their title pages or any front matter. It is therefore impossible to determine from which edition these pages might have come. (The JPEG images presented in the ARTFL edition (see below) are missing the same pages, which suggests a common, unknown, source.) In general, the Wikimedia Commons version does not identify the provenance for its images. This makes it a nice compendium of images, but useless as an actual scholarly source.

The Wikimedia Commons presentation also gathers together the individual page images into a series of single-volume files. This is good insofar as it gives one the ability to read the book as a book (albeit missing its front matter for vols. 9-17). Unfortunately, they chose the "DjVu" file format for this. The problem with DjVu is that it employs the "JB2" compression technique (a variation of the "JBIG2" technique). See the Note on the Problems with JB2/JBIG2, below for a discussion of why this renders the DjVu versions inherently unreliable.

The The French language "Wikisource" page on the Encyclopédie contains the text of the original edition in a transcribed format (in French, of course). This has many advantages for text searching, of course, but is in no case as satisfying or as accurate as actually reading the original printed text. Any transcription will, inevitably, place a layer of interpretation and potential error between you and the text. The original printing is lovely, and well worth reading as scanned images.

7. The 1959 Dover "Pictorial Encyclopedia"

[Physical book. Incomplete (by design). Good images. Limits on use.]

In 1959, Dover Publications, Inc. published a two-volume collection of plates from the Encyclopédie (with new captions in English):

Gillespie, Charles Coulston. A Diderot Pictorial Encyclopedia of Trades and Industry. (NY: Dover Publications, 1959)

There are positives and negatives to this volume.

On the positive side, the reproductions of the plates are very good (better than any of the digital editions). Although Dover claims ownership of the reprinted plates, it does allow limited use of them.

On the negative side, not all of the plates are included (and when they are included parts of them are sometimes silently omitted). There is no overall indication of what is or is not included; this is a picture-book, not an edition of the Encyclopédie. Although these plates were already nearly 200 years old in 1959, Dover claims ownership of these versions and allows only limited use.

These two volumes are currently (2014) out of print, but Dover reprinted them several times. Copies are still available at modest prices (copies are also offered to the unwary at outrageous prices - beware). They're really very nice to have.

8. Via Google Books

[Very bad. Incomplete, hard to find; scans of plates useless.]

Several volumes of various editions have been scanned by Google from various European libraries. However, their scanning is incomplete and in general it is exceedingly difficult to find a particular volume of any multivolume work in Google Books (this is of course a great irony, as Google began as a search engine). More significantly, Google is scanning for the purpose of OCR; the presentation of the texts is incidental to that. This means that for the volumes of plates, they do not fold out the plates when they scan. This, obviously, renders them useless.

Here are local copies of some of the volumes of the text from Google's versions. There are probably more available, but (again, ironically) finding them with Google is difficult. Note that in many cases pages were photographed imperfectly (e.g., in the middle of page turns) and cannot be read. I haven't cached any of the scanned volumes of the plates here, as I have no wish to embarass Google.

Note that the volume divisions of the Geneva and the Lausanne/Berne editions differ from those of the original (Paris/Neufchastel) edition.

Finally, note that if you pull apart the Google Books PDFs you discover that at a low level they are using a combination of JPEG2000 compression (which is fine) and JBIG2 compression (which, emphatically, is not). See the Note on the Problems with JB2/JBIG2, below for a discussion of how JBIG2 makes texts inherently unreliable.

9. Partial Source

(Other than the 1959 Dover volumes.)

9.1. US Library of Congress

Bruno, Leonard C. The tradition of technology : Landmarks of Western technology. Washington, D.C. : Library of Congress, 1995, p. 222.

This is a lovely volume published by the US Library of Congress. As an official publication of the US federal government, it is in the public domain.

A scan from this printed book is online on the US Library of Congress' website, at: http://hdl.loc.gov/loc.pnp/cph.3c10399 Title: "Imprimerie en lettres, l'opération de la casse". Here it is (click on the image below for a lossless PNG conversion of the L of C TIFF):

[click image to view larger]

image link-to-diderot-imprimerie-en-lettres-plate-1-us-loc-3c10399u-sf0.jpg

About the images