tipped-in pages signatures (in text collation) milestone reference system collation recto verso

Encoding of the collation of the document, recording both printed signatures as they appear on the page using mw type="signature" and also an idealized signature sequence using milestone unit="sig".

If signatures are of interest, it is useful to capture both the printed signatures that appear on the page, and the idealized signature for each page (whether or not a printed signature appears). The printed signature is transcribed as it appears on the page using mw type="sig". The mw element should be placed where the printed signature occurs on the page (usually right before the catchword, if any, or before the pb element that marks the start of the following page).

The idealized signature is recorded as the value of the n attribute of a milestone unit="sig". The milestone element should be placed at the top of the page, immediately following the pb element for that page. It is up to the individual project to decide how to regularize the representation of signatures, but in the absence of some factors that might determine otherwise, we recommend the following approach, which reflects standard bibliographic practice as adapted to the medium of text encoding:

Every signature is recorded in a form which includes the letter or letters (or in rare cases the numbers) that identify the signature, the number of the leaf, and an indication of whether the page is a recto or verso, in the form A1r, C5v, etc.

Delimiters, punctuation, and spaces within the signature should be omitted in the n attribute, except where they serve to disambiguate two signatures. For instance, a text might have signatures marked A, B, B., C, where the B. signature might be a later insertion differentiated by the period from the regular B signature. In this case, the period would need to be retained in the idealized sequence to avoid ambiguity.

In cases where the printer has used U and V interchangeably (for instance, a series of signatures marked Uuu, Uvu2, Uvv3, Vuu4, or the like), the idealized sequence should eliminate this variation, and should record whatever sequence is implied by the general pattern of the other signatures. For instance, in a long text whose signatures run

A, B, C,...Z, Aa, Bb, Cc,...Uv, Ww

the Uv signature should be regularized to Uu. Remember however that the actual printed signature as recorded in fw should represent exactly what is printed on the page without regularization of any sort.

Special cases:

Numeric signatures: In cases where the signature is numeric (as in some texts printed in North America), the leaf number should be enclosed in parentheses in the form 1(1)r to avoid ambiguity.

Offcuts: in some 19th-century American books the signature sequence includes a signature with an asterisk, usually on the third leaf of the gathering, which marks the start of the offcut (a smaller piece of paper included in the gathering). This asterisk should be ignored in encoding the idealized signature sequence, so a printed sequence looking like this:

1, [blank], [blank], [blank], [blank],[blank], 1*

would be encoded with n on milestone like this:

1(1)r, 1(1)v, 1(2)r, 1(2)v, 1(3)r, 1(3)v (in other words, as if the asterisk were not there).

Tipped-in pages: the n value of milestone should refer to the signature of the facing page, e.g. facing A2r.

Preliminary material without signatures: we recommend following the procedure recommended by Bowers and Tanselle, which is to use the Greek letter pi (entity &pgr;) for unsigned preliminary sheets, except where some ordinary letter could be reasonably inferred as part of the existing sequence. Thus (using P for &pgr;, and brackets to designate the inferred signatures):

[P1r, P1v, P2r, P2v], A1r, A1v, A2r, A2v


[A1r, A1v, A2r, A2v], B1r, B1v, B2r, B2v, etc.

For unsigned signatures which are interpolated between regular signatures, or which appear at the end of the signature sequence, Bowers/Tanselle propose using the Greek letter chi (entity &khgr;); thus, using K for &khgr; and brackets to designate the inferred signatures:

A1r, A1v, A2r, A2v, [K1r, K1v, K2r, K2v], B1r, B1v, B2r, B2v, etc.

For texts with multiple signatures of the same letter, the second (or greater) sequence is given a numeric prefix to distinguish it: 2A1, 2A2, 2A3, 2A4, etc. Thus the following sequence of printed signatures:

A1, A2, A3, A4, A5, A6, A7, A8,

A1, A2, A3, A4, A5, A6, A7, A8,

B1, B2, B3, B4, B5, B6, B7, B8,

C1, C2, C3, C4, C5, C6, C7, C8, ...

would be regularized on the n attribute as

A1, A2, A3, A4, A5, A6, A7, A8,

2A1, 2A2, 2A3, 2A4, 2A5, 2A6, 2A7, 2A8,

B1, B2, B3, B4, B5, B6, B7, B8,

C1, C2, C3, C4, C5, C6, C7, C8, ...