The Physiognomy of Lavater’s Essays: False Imprints, “1789” and “1792”
As to that false appearance which appears to the reasoner . . . it is a delusion of
William Blake, Milton (1804[-08?] pl. 28, ll. 15-16
John Caspar Lavater’s attempt to systematize his observations on human physiognomy into a kind of science in his Essays on Physiognomy created great excitement among his contemporaries. As The Gentleman’s Magazine, LXXI (Feb. 1801): 184, commented:
In the enthusiasm with which they were studied and admired, they were thought as necessary in every family as even the Bible itself. A servant would, at one time, scarcely be hired till the descriptions and engravings of Lavater had been consulted, in careful comparison with the line and features of the young man’s or woman’s countenance.
The Captain of the “Beagle,” it is said, almost rejected the young Darwin because of the shape of his nose,2↤ 2 John Graham, “Lavater’s Physiognomy: A Checklist,” Papers of the Bibliographical Society of America 55 (1961): 297; this and the essay on Lavater’s Physiognomy in England are reprinted, apparently without change, in his Lavater’s Essays on Physiognomy: A Study in the History of Ideas (Berne, Frankfort am Main, Las Vegas, 1979). By 1810, there were 16 versions of the Physiognomy in German, 15 in French, 2 in the US, 2 in Russian, 1 in Dutch, 1 in Italian, 20 in English, and by the 1940s there had been 160 publications of it (298-99). Graham lists the Hunter folio editions as (1) 1789-98; (2) 1792[-98]; (3) 1789-1810, in which the second and third entries seem to me to be significantly inaccurate. On 27 March 1962, Graham wrote to me that the source of his information for the 1792[-98] edition was “Auction, 1945-1950” and for the 1789-1810 edition was the catalogues of the Boston Public Library and the John Rylands Library. and “in many places, where the study of human character from the face became an epidemic, the people went masked through the streets.”3↤ 3 Encyclopaedia Britannica, Eighth Edition (1853-60), sive Physiognomy, cited by Graham, “Lavater’s Physiognomy in England,” Journal of the History of Ideas 22 (1961): 562.
One of the handsomest works printed during the great age of the illustrated book in England was John Caspar Lavater’s Essays on Physiognomy, translated by Henry Hunter. It was “A sumptuous Edition”4↤ 4 William Thomas Lowndes, The Bibliographer’s Manual of English Literature, rev. Henry G. Bohn, Part V (London: Henry G. Bohn, 1860) 1321. issues in five folio volumes5↤ 5 The five volumes consist of Vol. I, Vol. II Parts i and ii, and Vol. III Parts i and ii, and the second Parts of Vol. II-III have no titlepages. with “more than eight hundred engravings . . . executed by, or under the inspection of Thomas Holloway,” and published by John Murray, by the translator, and by the supervising engraver in 41 Fascicles over 11 years, from January 1789 to March 1799. (As the three titlepages are dated 1789, 1792, and 1798, this edition is identified here as that of 1789-98.)
The First Edition (1789-98)
It was recorded on its titlepages as:
ESSAYS | ON | PHYSIOGNOMY, | DESIGNED TO PROMOTE | THE KNOWLEDGE AND THE LOVE OF MANKIND. | BY | JOHN CASPAR LAVATER, | CITIZEN OF ZURICH, AND MINISTER OF THE GOSPEL. | ILLUSTRATED BY MORE THAN | EIGHT HUNDRED ENGRAVINGS ACCURATELY COPIED; | AND SOME DUPLICATES ADDED FROM ORIGINALS. | EXECUTED BY, OR UNDER THE INSPECTION OF, | THOMAS HOLLOWAY. | TRANSLATED FROM THE FRENCH |BY| HENRY HUNTER, D. D. | MINISTER OF THE SCOTS CHURCH, LONDON-WALL.|-| GOD CREATED MAN AFTER HIS OWN IMAGE. | - | [3 vols. but bound in five parts] VOLUME I. | [Vignette] | LONDON: PRINTED FOR JOHN MURRAY, NO. 32, FLEET-STREET; H. HUNTER, D.D. CHARLES’S- | SQUARE; AND T. HOLLOWAY, NO. 11, BACHE’S-ROW, HOXTON. M DCC LXXXIX .| VOLUME II [bound in 2 parts]|. . .|. H. HUNTER, D.D. BETHNAL GREEN ROAD . . . MDCCXCII . <See illus. 1> VOLUME III [bound in 2 parts]. | [Vignette] | LONDON. PRINTED FOR MURRAY AND HIGHLEY, NO. 32, FLEETSTREET; H. HUNTER, D.D.| BETHNAL-GREEN ROAD . . . MDCCXCVIII .
The description on the titlepages of Vol. I (1789), Vol. II (1792), and Vol. III (1798) of “MORE THAN EIGHT HUNDRED ENGRAVINGS” was, however, greatly exaggerated, for, when the copperplates were sold in 1818 (see below), there were only 537 of them. However, the exaggeration was apparently not noticed until 1799, when a somewhat lame note on “Errata” was added to the Directions to the Binder (bound at the end of Vol. III, Part ii and generously pointed out to me by Mary Lynn Johnson):
In the Title pages, dele[te] ‘Upwards of Eight Hundred.” This error was not discovered before the titles were printed off, and it was thought better to let it pass than deprive the purchasers of fine impressions of the Vignettes which they contain.begin page 17 | ↑ back to top
From its first appearance, the work was admired for the beauty of the printing and the engravings. It was described in The Monthly Magazine, VIII (Jan. 1800): 988, as “the finest print book that ever appeared in this or any other country,” and T. F. Dibdin said that “the copper plates . . . have scarcely been equalled.”6↤ 6 T. F. Dibdin, The Bibliographical Decameron (London: The Author, 1817) 398, 399. It generated additional interest when the leaders of the Blake revival noticed that it had prints by William Blake. Blake’s part in it was recorded by, inter alia, William Haines in 1863,7↤ 7 Alexander Gilchrist, Life of William Blake, “Pictor Ignotus” (London: Macmillan, 1863) 2: 260 (catalogue of Blake’s engravings by William Haines, recording only two of the four Blake engravings; for the identity of the compiler, see Blake Books  815). A. G. B. Russell in 1906,8↤ 8 Archibald G. B. Russell, The Engravings of William Blake (Boston & NY: Houghton Mifflin, 1912) 152-53, recording three Blake engravings. Geoffrey Keynes in 1921,9↤ 9 Geoffrey Keynes, A Bibliography of William Blake (NY: The Grolier Club, 1921) 233-34, recording three Blake plates. and G. E. Bentley, Jr. & Martin K. Nurmi in 1964.10↤ 10 G. E. Bentley, Jr. & Martin K. Nurmi, A Blake Bibliography (Minneapolis: University of Minnesota Press, 1964) 137-38, recording three Blake plates. The fourth plate was pointed out in “A ‘New’ Blake Engraving in Lavater’s Physiognomy,” Blake Newsletter 6 (1972): 48-49.
The work suffered in a number of ways as it passed through the press. For one thing, the original publisher, John Murray [I] died on 6 November 1793, not long after the titlepage to the second volume had been issued (1792). During the minority of his son John Murray II (1778-1843) and for a few years beyond, the firm was knows as Murray & Highley (1793-1803), and the only Lavater titlepage issued during this period (for Volume III) bears the names of Murray & Highley.
For another, the cost of the engravings, the major expense of the work, far exceeded the receipts from the subscriptions for the fascicles, at least for a time. The cost of engraving the copperplates must have been from £3 for small, simple outlines to more than £100 for large, ambitious plates. If the 174 large plates averaged only £40 each (an exceedingly modest estimate), they alone would have cost almost £7,000. When John Murray [I] died in November 1793, perhaps 60% of the engravings had been finished,11↤ 11 Most of the plates have no imprint, so it is very difficult to ascertain how many had been finished by November 1793. but the firm still owed thousands of pounds to the engravers.12↤ 12 In “the publication of Lavater’s work on Physiognomy, the engraving of the plates caused the principal part of the loss. The executors put the case to arbitration, and were eventually compelled to pay out of the estate the sum of £3,900” (Samuel Smiles, A Publisher and his Friends: Memoir and Correspondence of the Late John Murray, with an Account of the Rise and Progress of the House, 1768-1843 [London: John Murray, 1891] 1: 28).begin page 18 | ↑ back to top
Despite these difficulties, the work was finally finished in 1799, and a very handsome work it was. This was the only visage of the work with Blake’s prints known, or at least recorded, for many years.
Other Editions (“1792” and “1810”)
However, there are two other folio editions of Lavater’s Essays on Physiognomy as translated by Hunter, whose features are as handsome as those of the original.13↤ 13 According to Lowndes, Part V, 1321, “Some copies were subsequently issued with the name of Stockdale as publisher [i.e., in 1810]; these are inferior in impression to the first edition. . . .” However, the type, impression, and plates in the 1810 edition appear to me to be quite worthy of the edition of 1789-98. Only in 1977, more than a century after Blake’s part in Lavater’s Physiognomy was first noticed, or at least recorded, was it remarked in print that there are two other editions dated 1792 and 1810 of Lavater’s Essays on Physiognomy bearing Blake’s prints. And even then it was not noticed how odd it is to have a complete edition dated 1792 finished before the first edition of 1789-98 was accomplished. The somewhat feeble conclusion offered at the time was the “The 1792 issue was presumably for late subscribers, with the titlepage of Volume III ante-dated” (Blake Books  594).
The chief printed distinctions between the first edition of the Hunter translation and its successors are little more than in the dates and the publishers on the titlepage (John Murray in 1789-92-98, John Murray in 1792, Thomas Stockdale in 1810). In general, the “1792” edition is more like the first edition of 1789-98 than is that of 1810, as one might expect. The titlepages of the first Hunter edition (see illus. 1) and those of “1792” (see illus. 2) claim to have “UPWARDS OF EIGHT HUNDRED PLATES,” whereas that dated[e] 1810 omits “UPWARDS OF EIGHT HUNDRED PLATES” <see illus. 3>.14↤ 14 G. E. Bentley, Jr., Blake Books (Oxford: Clarendon Press, 1977): 593-95.
There seemed, then, to be three folio editions of the Hunter translation of Lavater’s Essays on Physiognomy, the first issued in fascicles in 1789-99 with titlepages dated 1789 (Vol. I), 1792 (Vol. II), and 1798 (Vol. III), the second in 1792, and the third in 1810. And the Blake prints did not seem to change significantly in any of them.15↤ 15 “The Blake plates are not changed” (Blake Books 594); see also Robert N. Essick, William Blake’s Commercial Book Illustrations (Oxford: Clarendon Press, 1991) 41-42. At last our understanding of Lavater’s book seemed to have stabilized.
“I discern thee other then thou seem’st.” Milton, Paradise Regained, I, 24816↤ 16 The phrase is used in James T. Wills, “‘For I discern thee Other Then Thou Seem’st’: An Extra Illustration for Blake’s Paradise Regained Series,” Blake Studies 8 (1980): 109-19.
Then, when the last changes in the text of Blake Books Supplement were being sent off to the press in the autumn of 1993, I received a puzzled letter from Stewart Naunton (then unknown to me) who had been given a set of the 1792 Lavater, and, in making a more meticulous examination of it than any professional bibliographer apparently had done, Naunton had discovered that what he had was a rogue. Though the titlepage plainly said 1792 (or at least MDCCXCII), the leaves had watermarks of “1806,” “1812,” and “1817”—and none of the watermarks exhibited a date consonant with the “1792” on the titlepage.
This was very worrying. Was the “1792” set described an unique rogue?—but how could just one copy be printed thus? Were there other “1792” sets with anomalous water-marks? Perhaps all “1792” sets were in disguise like the one described to me, bastards of c. 1818? (just a few years after the legitimate 1810 edition) masquerading as legitimate heirs of the 1789-98 edition.
Further, the “1792” set reported to me used paper in a very curious pattern, rarely exhibiting more than a few sheets of the same paper together at a time. This seemed a remarkably irregular, not the say inefficient, way to use paper. Did other sets duplicate these oddities?begin page 19 | ↑ back to top
I immediately wrote off to all the institutions which are recorded in the Eighteenth-Century Short Title Catalogue and the National Union Catalog as having sets of the “1792” edition17↤ 17 The libraries of the University of Chicago, Cooper-Hewitt Museum, Duke University, Emory University, the University of Iowa (reported by Mary Lynn Johnson), the University of Kentucky, Liverpool University, McGill University, Newcastle University, Oregon State University, and Princeton University are recorded as having the “1792” set of Lavater. I am grateful for the information generously supplied by the librarians of the University of Chicago, the Cooper-Hewitt Museum, Duke University, Emory University, the University of Kentucky, Stewart Naunton, Oregon State University, and Princeton University. to enquire about their watermarks.
Of course making a record of all the watermarks in five fat volumes is immensely laborious—and a feat which apparently no bibliographer had previously attempted. I expected to receive from the libraries to which I had written no more than a sampling of watermarks to ascertain whether they were or were not consonant with the “1792” date on the titlepage, and this of course was what I received from most institutions which replied. But J. Samuel Hammond, Rare Book Librarian at Duke University, recorded every watermark in the first volume, including laid or wove paper without letters or numbers, and Karen Herbaugh, Student Assistant at the Kerr Library of Oregon State University, provided me with a list of watermarks on every leaf in all five volumes. I am deeply in the debt of all the librarians who responded so generously to my queries and especially to Hammond and Herbaugh.
The results are yet more curious than my first warning had led me to expect. In the first place, a very surprising variety of paper was used. The text is on paper watermarked “1 7” (4.5 cm apart, in the middle of the margin), “1801,” “1804,” “1806,” “1809,” “1817”18↤ 18 Some watermarks of “181” are reported, but I suspect that this represents “1817,” the last digit of which is often hard to see or decipher. Occasionally “S” or “II” is visible on the same leaf as “1817.” plus “II” or “TH,” “S,” “WL,” “95 LEPARD,” and “LEPARD” (all wove paper except “LEPARD,” which has chain lines), and the prints are on paper watermarked “1806,” “1812,” “1813,” “1814,” “LEPARD,” “SMITH,” “W SPEAR 1815,” and “JWHATM,” and “1794 | J WHATMAN.” More than one of these post-1800 watermarks appear in sets of the “1792” edition in the libraries of the University of Chicago, the Cooper-Hewitt Museum, Duke University, Emory University, the University of Iowa, the University of Kentucky, Liverpool University, Oregon State University, Princeton University, and Stewart Naunton, and there is every reason to expect that all sets have a similar pattern of watermarks. And none of begin page 20 | ↑ back to top the paper watermarked “1801,” “1804,” “1806,” “1812,” “1813,” “1814,” and “1817 can have been used at the time of the 1792 date on the titlepage.
In the genuine first edition of 1789-98, the only watermarks in the text are “LEPARD,” “Lepard,” and “LW,” with occasionally “17” and “95,” all on laid paper (at least in my own set and that of the University of Toronto). In the 1810 edition, the only watermark in the text of my set is “1806” plus “II” once in every gathering of wove paper, and these watermarks too recur in the “1792” edition.
The second oddity is the way in which the paper was used. In the Princeton set of “1792,” which I went down to examine myself, successive gatherings are often on distinctly different paper. For instance, in Volume I, gatherings P-T are on wove paper with “1804,” “1806,” “S,” “1817,” and without watermark, and laid paper without watermark. This is a very peculiar way to print, with a different kind of paper in the stack for each gathering.
But in fact, the situation is even odder than this, for apparently there were no stacks of “1804” and “1806” and “1817” paper. Instead, each stack of sheets was apparently of mixed watermarks for each gathering. For instance, the paper for four successive gatherings (G-K) in Volume I of the “1792” sets in Duke, Oregon State, and Princeton looked like this:
And other gatherings show similar absence of pattern. Clearly the stack of sheets for each gathering were very mixed indeed as to watermark.
Such a mixed stock of paper would have been most unusual even in the shop of a small-scale jobbing printer. After all, one buys paper by the ream or by the pound all with the same watermark, and to mix the papers thus would have been troublesome. Fairly plainly, these are odd sheets left over from other jobs and mixed higgledy-piggledy to save space. But it is very surprising to find paper thus mixed up in the work of a fine printer like Bensley, apparently the printer of all three editions.
The question of who printed Lavater’s Physiognomy, or rather of what types were used, is of some significance. A prospectus for the work of June 1787 said that the type was to be cast on purpose for it, and “The Letter-Press will be executed in the most superb Manner.”19↤ 19 Collectanea: or, A Collection of Advertisements and Paragraphs from the Newspapers relating to various Subjects (Printed at Strawberry-Hill By Thomas Kirgate, for the Collector, Daniel Lysons), f. 81v <British Library: 1881 b 6>. Dibdin (ut supra) said that the five-volume Lavater (1789) was Bensley’s “earliest attempt at fine printing,” and the high quality of the type and printing seems to my typographically unsophisticated eye to be worthy of him. Further, as the titlepages of 1789-92-98, 1810, and 1792 indicated (see illus. 1-3), the same type seems to have been used for all three editions. Whoever the compositors were, they seem to have been using the same type in all three editions. And it seems very odd that such careful composition and printing should have been associated with such a casual use of paper.
The pattern of watermarks for all three editions may be seen below:
↤ 20 The 1802 watermark is found at Vol. II, Part i at 3Y2 and Vol. III, Part ii at 3R2 and 3U2 in the Huntington set, which has binder’s flyleaves watermarked 1801. ↤ 21 The “LEPARD” watermark appears in two different positions in laid paper, one parallel to the chain lines and one at right angles to them, as Detlef Dörrbecker pointed out to me. ↤ 22 Mary Lynn Johnson and John Grant report that the lettering with “1784” in the University of Iowa copy is exceedingly difficult to decipher and that other plates there have different indecipherable watermarks. ↤ 23 N.b. With all copies save those I have examined myself for the purpose (i.e., GEB, Huntington, Princeton, and Toronto), I cannot be sure whether the watermark is “LEPARD” or “Lepard” or whether the paper is wove or laid.
|95||laid paper||180220||wove paper|
|LEPARD21||laid paper||Adviune[?] 178422|
|LEPARD 95||laid paper||LW||wove paper|
|WL 94||laid paper||J Whatman | 1794||wove paper|
|95||wove paper||1812||wove paper|
|1817||wove paper||W. SPEAR 1815|
|II or TH||wove paper||J WHATMAN|
|LEPARD||laid paper||1794| JWHATMAN|
Note that all copies of a text-sheet must be printed at the same time (i.e., in one more or less continuous print-run), so that the type can then be distributed and used to set and print other sheets. If more copies of the text are needed after the type has been redistributed, for instance when a decision is made to increase the size of the print-run, the text must be composed and printed again.
However, the plates can be printed as the need arises, since the copperplates are scarcely affected by the act of printing. Further, it is far more expensive and laborious to print plates than to print type-set text, and consequently it is sensible to pull only as many prints as are manifestly needed at the moment, and to wait until sales demonstrate the need and provide the capital to print more copies. If, for instance, one had 500 subscribers and hoped to sell a thousand copies altogether, one might print a thousand copies of the text but pull only 600 copies of the prints, assuming that more copies of the prints might be pulled as demand warranted.
As a consequence, we should expect the text of all copies of an edition to reflect the date of printing and first issuing them; in the case of the 41 Fascicles of the first edition of Lavater’s Physiognomy, this would be 1789-99—and the watermarks do reflect this, or at least do not contradict it. However, the last copies sold might well include prints pulled long after the last of the text had been printed, and in at least one copy of the first edition of Lavater’s Physiognomy this is the case. In the Huntington set, three of the prints are paper watermarked “1802” (Vol. II, Part i, at 3Y2, and Vol. III, Part II, at 3R2 and 3U2), and the work was probably bound not long after the “1801” watermark in the binder’s flyleaves.
This “1792” edition, then cannot be earlier than 1817, the latest date in the watermarked paper in all copies examined for watermarks, 25 years after the date on the titlepage. But did the compositors for the “1792” edition follow the text of the first edition of 1789-98, as they followed its titlepages (though dating them all “1792”), or did they follow that of 1810?
The printing is very careful and handsome in all three printings, but clearly they are separate editions rather than re-issues of the same sheets or re-impositions of the same type. The watermarks alone indicate that the third edition of “1792,” with watermarks of 1817, cannot be the same sheets as in the first edition of 1789-98, with text-sheets watermarked 1794 and 1795.begin page 22 | ↑ back to top
Page-for-page and line-for-line the three editions are virtually identical in type, but there are significant differences in spacing in the second edition of 1810 and the third edition of “1792,” indicating that they are different settings of type rather than reprinted from standing type.24↤ 24 In any case, it would have been commercial folly to keep the type for a five-volume folio work standing for a period of almost 30 years. Of course it could be done, but what would be the incentive, especially when the engravings, the raison d’etre of the edition, had been sold to another publisher?
The precision with which the text was reset is, to me, astonishing. Taking as a test Vol. I, p. 31 (sig. L1r) of the three editions of 1789, 1810 and “1792” (1818?), the ordinary tests demonstrate that the house style produced almost identical results for pages set some 30 years apart. For instance, if one lays a straight line, such as a ruler, across the page between two arbitrarily-chosen points, say the top left and bottom right corners of the text, the line intersects identical fragments of text in each edition, though they are printed on paper watermarked 1794, 1806, and 1817. Minute details of composition, such as the large space after the opening double quotations marks (“), the small one before a closing set, and the space before a semicolon, are uniform throughout all three editions. Since the compositor sees only one line as he loads his stick with type, the spacing within a line may be controlled very precisely, but the relationship of line to line is far more difficult to control and is far more a matter of chance. The near-identity of spacing throughout the page in each edition demonstrates an extraordinary uniformity of composing techniques and suggests to me that all three pages were set by the same rules of composition and probably by the same printing-house.
Similarly, ligatures (e.g., “fi,” “si,” “sh,” “st,” and “ct”) are the same in each edition, and all three editions use the long “s,” though in 1810 and 1818 this was an anachronism.
The clearest of the tiny differences between the three editions of Vol. I, p. 31 are between the second edition of 1810 as compared to the first and third editions of 1789 and “1792” (1818?) (see illus. 4-5). For instance, in the last word of text (“sign?”), there is a substantial space before the query in 1789 and “1792” (1818?), putting the query under the “g” of “energy” above it, while in 1810 the space before the query is markedly slimmer, and the query is beneath the “r” of “energy.” Yet more strikingly, “energy” is followed by a colon in 1789 and “1792” (1818?) but by a semi-colon in 1810.
Similarly, in Vol. I, p. 279 (sig. E1r) in the second edition (1810), in the second series of asterisks the first asterisk is under the comma of “upon,” but in the third edition (“1792”) it is under the “n.”25↤ 25 Note that the text-sheets watermarked “17 LEPARD 95” and “LEPARD” in the third edition (1818?) can scarcely be from the 1810 printing, for all the sheets so watermarked (e.g., Vol. I, sig. B) in one set (e.g., Oregon State and Princeton) have much later watermarks in other copies (e.g., “1809” in Duke). It is just possible that the prints watermarked “LEPARD,” “J WHATM,” and “1794 J Whatman” were from the early printing, but the fact that most of the other plate-watermarks are dated 1806 and later makes this seem unlikely. In the first edition (1789-98), Vol. II, Part i, pp. 143-50 (Nn2-Pp1) are numbered correctly; in both the second edition (1810) and the third edition (1818?) they are mis-numbered as pp. 135-42. Since the third edition marked “1792” cannot have been printed earlier than 1817 (as shown by its watermarks), it compositors must have copied the text of the second edition (1810) rather than (as the titlepages would lead one to expect) the first edition of 1789-98.
The differences between 1789 and “1792” (1818?) are even smaller—and never, to me definitive (see illus. 4-5). I can persuade myself that there are minute, scarcely measurable differences of spacing, but a re-examination of the same spacing usually leaves doubts as to whether there are clear differences. The text in 1818? must have been re-set when it was printed on paper of 1817, some 30 years after the original type was distributed, but the facsimile is so close as to make it, to me, virtually undetectable.
There can be little doubt that the 1818? edition was created with the intention to deceive, for though the text itself derives inconspicuously from that of 1810, the titlepages copy the wording of the first edition of 1789-98, with “UPWARDS OF EIGHT HUNDRED ENGRAVINGS”, rather than that of 1810, which omits reference to the eight hundred.
I would be agreeable if these perplexities could be enlightened by reference to the records of the publisher, but unfortunately the firm of Murray, whose name appears prominently on the editions of 1789-98 and of “1792” and which still survives today, has no trace of Lavater’s Essays on Physiognomy.26↤ 26 As Sir John Murray wrote to me on 10 February 1961.
The evidence seems unambiguous that the “1792” edition of Lavater’s Essays on Physiognomy is a line-for-line reprint of the 1810 edition printed in or after 1817. But what was the incentive to print another edition so soon after the edition of 1810? If it took from 1789 until about 1810 to exhaust the stocks of the first edition (1789-98), it is unlikely that only about seven years would have sufficed to exhaust the stocks of the 1810 edition.
The answer may lie in the auction by Mr. Saunders of the 537 copperplates for Lavater’s Essays on Physiognomy with Stockdale’s stock which was announced in a prospectus of 2 January 1818.27↤ 27 British Library: 11902 bbb 23, Vol. II. I have not seen or traced the auction-catalogue itself. Perhaps the purchaser attempted to capitalize promptly upon his acquisition of the copperplates by publishing a new edition of Essays on Physiognomy. And if the new owner did not also own the copyright to Hunter’s translation of Lavater’s text, it may have seemed discrete to publish the work with a false imprint. In that case, the “1792” edition is not only a bastard but a piracy.begin page 23 | ↑ back to top
This of course does not identify the fraudulent publisher of the “1792” Lavater, but it does help to suggest an occasion and a motive for publishing the work with a false imprint 25 years or more after its apparent date.
Sheet-by-sheet examination of watermarks is extraordinarily tedious, keeping tallies of them is remarkably intricate, and in most cases what is learned is of trifling significance—the paper, even if it is dated, is almost invariably consistent with the date on the titlepage.28↤ 28 Instructive exceptions are John Gay’s Fables (Stockdale, 1793, also issued with watermarks of c. 1811), John Flaxman’s designs to Homer’s Illiad (1793), Odyssey (1793), Dante (1793), and Aeschylus (1795) which bear watermarks of a number of different countries and dates, and most of William Blake’s works in illuminated printing which bear watermarks dated as much as 43 years later than those on the titlepages. Most bibliographers and editors probably check only a few sample sheets for watermarks, for “In doubtful questions ’tis the safest way / To learn what unsuspected ancients say,”29↤ 29 Dryden, Religio Laici. and leave it at that.
The danger is when the ancients are wrongfully unsuspected. Only when there is reason to be suspicious will a full-scale examination be made—or at least that was plainly the case with the credulous W. T. Lowndes, William Haines, A. G. B. Russell, Geoffrey Keynes, G. E. Bentley, Jr., and R. N. Essick. Doubtless one lesson to be drawn from this problem is that we should all be somewhat more suspicious than a really nice nature would ordinarily permit.
But the problem with Lavater’s Essays on Physiognomy is even more intricate than we have yet seen. We have thus far established that there were three editions, the first honestly dated 1789-98, the second honestly dated 1810, and the third marked 1792 but really of 1817 or later. Two or three libraries30↤ 30 E.g., Boston Public Library, Rylands Library of the University of Manchester, and the University of Toronto. claim to have editions of 1789-1810, but these are clearly mixed sets of the first and second editions.
Or so I thought until I examined the University of Toronto set of the 1789-1810 edition. To my considerable dismay, I found that Vol. I (“1789”) and Vol. II (“1792”) have watermarks of “1804” and “1806”! (Vol. III  also has watermarks of “1806,” but of course these are not anachronistic.) Further, the titlepage and text have been reset with very remarkable fidelity to the first edition. For instance, on the titlepage of Vol. I the “U” of “HUNTER” is under the “B” of “BY” in the first edition (see illus. 6) but under the “Y” in the edition watermarked “1806” (see illus. 7). Since the watermarks of Vol. I (“1789”) and Vol. II (“1792”) with the 1810 Vol. III are the same as in the sets in which all three titlepages are dated 1810, it seems fairly plain that they are part of the 1810 edition and that in 1810 two titlepages each were printed for Vol. I (dated “1789” and 1810) and Vol. II (dated “1792” and 1810).
This give us for Lavater’s Physiognomy translated by Henry Hunter three Volume I titlepages dated 1789 (one honest, one of 1810, and one of 1817), three Volume II titlepages dated 1792 (one honest, one of 1810, and one of 1817), and three Volume III titlepages dated 1798 (honest), 1792 (i.e., 1818?), and 1810 (honest).
In the copies I have examined or which have been examined for me, sets with titlepages of 1789-98 and 1810 are apparently honest, while those in which the three volumes are dated 1792 or 1789-1810 are fraudulent (though Vol. III dated 1810 is apparently always honest).
But it does not bear thinking of how many of the apparently honest sets of 1789-98 are mixed sets, with Vol. I-II fraudulent and Vol. III honest. And of course there may be a Vol. III marked 1798 which is also fraudulent.
The final tally, or at any rate the tally to date (23 August 1995), for Lavater’s Essays on Physiognomy translated by Hunter with the plates supervised by Holloway is:
|Edition||Volume I||Volume II||Volume III|
|Second (1810)||1789 or 1810||1792 or 1810||1810|
And that’s quite enough intricacy to be going on with, it seems to me. I at any rate will endeavor in the future to be as suspicious as a really nice nature will permit and will glance at watermarks regularly—but I bet nothing as devious and intricate as Lavater’s Essays on Physiognomy turns up again.