432 Proceedings of the Boijal Irish AcaSemy. 



four different exemplars.* Each of these may have had headings which were 

 not transferred to L ; but it is certain that they were all far from complete. 

 The number derived from each source is approximately as follows : L'49, 

 \I 57, P47, L* 13. For the rest, B has 124 headings, F123, H107, M99, Q9S, 

 P82, E77, T9, Zo. K was probably not well supplied with headings; for of 

 the eleven psalms in regard of which its evidence is available (xl, xlvi, Ixiii, 

 Ixviii, Ixxviii, Ixxx, Ixxxi, Ixxxii, Ixxxiii, cii, cv), only six have them. The 

 Salaberga Psalter,|| which is of somewhat earlier date, is probably in like case. 

 The published facsimiles show the beginnings of ten psalms, of which no 

 more than four (probably only three) have headings. It is clear therefore 

 that complete sets were comparatively rare, and that headings tended to 

 disappear in the course of transmission. There was also a tendency to 

 abbreviate headings, which becomes more evident when the various series are 

 classified according to their affinities.^ 



Table I reveals a remarkable variety among the different sets of headings. 

 Five sets stand so far apart from one another and from the rest that they 

 may be regarded as representing different types — DELANO ; and a sixth has 

 marked characteristics of its own — L'. The remaining sixteen — ABCFGHK 

 DL'MPQESTZ — have a general resemblance to each other, and over against 

 those just . mentioned may be regarded as a class conforming to a single 

 type. But even among them there are clearly defined groups. It has 

 been shown already that AR are derived from a single, not remote 

 ancestor, a; and C is obviously akin to them. FHQ form another group. 

 They are in agreement, exact or approximate, against all others (except G) nine 

 times: Pss. xxiv, xxxvii {salutis), xliii, xlvi (ji. cqwstolorum est), 1, Ixxii, cv, 

 cvi, cxvii. FH agree against the rest ten times (in six of which Q gives no 

 evidence) : Pss. ii, ix, xii, xxix, Ixiv, xcviii, ci, cxviii, dakth, hcth, iod. 

 FQ twice (H giving no evidence in one case) : Pss. li, lii. HQ nine times 

 (in one of which F fails us) : Pss. xviii, xlii, liv, Ixvii, Ixxi, Ixxxix, xciii, xcv, 

 ciii. This suggests that FHQ are descended from a common ancestor, which 

 we may call (/>. And the hypothesis is confirmed by a consideration of the 

 relation between the Mss. in other cases. All the facts are explicable by 

 derivation from ^ except the following : once HQ seem to have been 



* Headings in gi-oen ink are corrected in bine : Pss. 1, xcv, cl. Headings in black 

 ink are corrected in blue (Ps. cxlii) or green (Ps. cxxvi). Thus the order of date .seems 

 to be black (L^), green (L-), blue (L'). But, on the other hand, some blue headings 

 seem to be corrected in green : P.ss. cxxvi, cxxxiv, cxlix. This apparent conflict of 

 evidence might be removed by a careful study of the MS., which I have not been able to 

 undertake. It seems impossible to fix the place of the violet headings (L^) among the 

 rest, though once violet (L^) is corrected in blue (L'). || See p. 413, note '. 



§£.3., Table I shows that HFQ often abbreviate the readings of a. 



