s form a sub-class of this Type.]

57. There is another curious circumstance about the MSS. of this type,
viz., that they clearly divide into two distinct recensions, of which both
have so many peculiarities and errors in common that they must necessarily
have been both derived from _one_ modification of the original text,
whilst at the same time there are such differences between the two as
cannot be set down to the accidents of transcription. Pauthier's MSS. A
and B (Nos. 16 and 15 of the List in App. F) form one of these
subdivisions: his C (No. 17 of List), Bern (No. 56), and Oxford (No. 6),
the other. Between A and B the differences are only such as seem
constantly to have arisen from the whims of transcribers or their
dialectic peculiarities. But between A and B on the one side, and C on the
other, the differences are much greater. The readings of proper names in C
are often superior, sometimes worse; but in the latter half of the work
especially it contains a number of substantial passages[7] which are to be
found in the G. T., but are altogether absent from the MSS. A and B;
whilst in one case at least (the history of the Siege of Saianfu, vol. ii.
p. 159) it diverges considerably from the G. T. _as well_ as from A and
B.[8]

I gather from the facts that the MS. C represents an older form of the
work than A and B. I should judge that the latter had been derived from
that older form, but intentionally modified from it. And as it is the MS.
C, with its copy at Bern, that alone presents the certificate of
derivation from the Book given to the Sieur de Cepoy, there can be no
doubt that it is the true representative of that recension.

[Sidenote: Third; Friar Pipino's Latin.]

58. III. The next Type of Text is that found in Friar Pipino's Latin
version. It is the type of which MSS. are by far the most numerous. In it
condensation and curtailment are carried a good deal further than in Type
II. The work is also divided into three Books. But this division does not
seem to have originated with Pipino, as we find it in the ruder and
perhaps older Latin version of which we have already spoken under Type I.
And we have demonstrated that this ruder Latin is a translation from an
Italian copy. It is probable therefore that an Italian version similarly
divided was the common source of what we call the Geographic Latin and of
Pipino's more condensed version.[9]

Pipino's version appears to have been executed in the later y