:: :: University of Virginia Library

Within the first several years the variation is of such complexity that it is expedient to consider separately two kinds of evidence. Accordingly the tables pertain, IV to issue-sequences immediately recognized by title-points, V to supplementary data identifying the state of numbers now lacking titles.

Table IV successively records, for each number, three features of the title:

1. A blank for original edition, a hyphen for an unlabelled reprint, a "2", "3" or "N" etc. for "Second", "Third", "New" or other edition so identified.
2. Within parentheses, a letter indicating the woodblock (as described in Table II), together with a superscript figure for cuts in later state.
3. After parentheses, a number identifying the imprint (as defined in Table III), together with a letter for variant states and, when this is given, a subscript figure representing the number of volumes cited, in the imprint, as then available.

Among these varying features the imprint alone may indicate earliest dates of issue. Thus, in the absence (a) of a woodblock, it would appear that imprint 3, used originally from March to May 1731, establishes within that period the issue of second editions similarly imprinted. So also imprint 5, in July-August 1731, determines the issue of second and third editions of identical reading.

Once the earliest block (b) appears, two criteria are available, and time of issue therefore established, usually, by the conjunction of these two. Hence the fourth edition earliest number could not have appeared before February 1732 since the cut (b), though extending from September 1731 to April 1732, is not associated with imprint 9 before February.

Upon the juncture of two or more points, however, any discrepant factor must necessarily fix the date. Thus while the two "fifth" editions earliest number would seem to be of the sequence (d)-(e), displaying in order the "dog" (1732) and "equestrian" (1733) block, this order is reversed by imprint sequence 15 and 22, occurring respectively in 1733 and 1753. Thus again the counterfeit "sixth" edition, though also seemingly

of the year 1742 (when 11 volumes were available), and also bearing imprint 20 (one appropriate to that time), is betrayed by its woodcut (h2) as of an issue not appearing in any original edition before 1781—and, as already suggested (note 21), not evident in any reprinting before 1786.

In checking the variants of a certain number it will be noted that, excepting the first two, the "edition" sequence, if any, is quite disorderly. For July 1731 early accumulated sets exhibit three different "first" editions: A-F the original setting (a) 5; G-H a second setting -(b)9, prepared apparently about the same time as the G-H fourth edition first number [4(b)9]; and I-J a third setting -(e) 17, issued it would seem about a year after I-J fifth edition first number [5 (e) 15]. Then in K-P comes a typographical arrangement appropriate only for 1746 [(e1) 2015], now in the July reprint improperly labelled "fifth" edition; in February, however, correctly described as "sixth" edition; but then, when used three years later for an August number [-(e1) 2018], passed off as another "first" edition—here as the second of three "firsts" following an actual second edition. Apparently in some instances the reprint arrangement is that of the title then being used for first editions; in others, an imitation of the copy now being reprinted; in others still, a combination arbitrarily devised.

Since these irregularities diminish with each passing year, it is convenient to reduce the 16 copy-sequences, in 1732 to 7, in 1733 to 4, and in later years (as noted Table VI) to 1. Throughout, however, notes have been appended to signify other peculiarities either in form or setting.