With the exception of the final quire, the manuscript comprises entirely of regular gatherings of twelve leaves. The consistency in the copying of the text, its presentation and that of accompanying paratext, and in the execution of decorated initials, all points to the manuscript having been conceived and produced as a whole. However, as H.L. Pink observed in his unpublished description (see Source), there is at the end of several quires the word 'quaternus', written in the very bottom right-hand corner of the page. These have for the most part been lost when the edges were cropped during binding, however enough survives at the end of Quire 16 not only to confirm that these are quire signatures; since Quire 16 is 'quaternus iiij^us', the numbering must have restarted at Quire 13, where a new text begins after three blank leaves at the end of Quire 12. It seems, then, that the quires at least were prepared separately, leading one to conclude that these were two codicological units. In the end, they may have been written on and decorated by the same scribe and the same artist, or by two or more hands capable of imitating one another's work, and the presence of medieval pagination (rather unusually) nevertheless confirms that they were soon bound together.

Both M.R. James and H.L. Pink thought Quire 20 was, like all its predecessors, another gathering of twelve leaves, but from which the last two leaves were excised. That may indeed have been the case. However, the formula presented here is more circumspect, in light of ambiguities in the material evidence. The leaves have signatures, though not in a form seen anywhere else in the volume ('i-vi', written in ink). At face value, these seem to confirm James and Pink's interpretation, but for the presence of two stubs between ff. 231 and 232. Comparison with a copy of De natura rerum in Cambridge, Trinity College, MS R.16.4 (at f. 219v) , confirms that no text is missing. If two leaves were cancelled here, the quire would originally have been of sixteen leaves, from which four leaves are absent at the end, not two. In a manuscript of otherwise entirely uniform structure, this seems unlikely, and in any case is contradicted by the medieval leaf signatures. It is possible that these were added later, though, in order to prevent any confusion at the point of binding the one irregular quire in the manuscript. Unfortunately, the leaves are sewn in at the head- and tailbands, and conservation repairs along the spinefolds of ff. 238 and 239, making it impossible to see how leaves in each half of the quire connect with one another and to what these two stubs are attached. Hence, this collation formula records simply the presence of ten leaves.

• Quires 1-12¹² (ff. 2-145) ;
• Quires 13-19¹² (ff. 146-229)
• Quire 20^ten (ff. 230-239

1-12¹² ; 13-19¹² 20^ten