$ 54. Pierias C, plerias ABE, proelias P, pleridas V. Here it would seem that a must have had plerias, variously corrupted in PV; C is an easy emendation.

§ 65. consulantne di rebus humanis C, for di the other мss have de, which is a very natural corruption of original di, and was probably the reading of a, corrected by C.

§ 72. quod sumas, quanto dissipes libentius CE; for dissipes ABPV' have dissipis, V dissipas. Here too I think it is easiest to regard C's reading as a correction of dissipis.

$74. de fide mala tutele (for -lae) C, de fide fidem alatat utile A, de fide allata tutelae B, de fide mala at utile PV, de fide mala tot utiles E. Here A must have arisen from a wrong division of words (de fidem alat utele) and a double dittographia of fide and at: perhaps in a the line ended with at. I think it probable that here too C's reading is an emendation, as B's certainly is.

§ 83. grave C, gravem all others. Probably amiculum was mistaken for the diminutive of amicus and so made masculine. C's reading I take to be an emendation.

§ 86. cuipiam nocuit C, quipiam n. A'B, quippiam n. A'V (nothing said of E and P). Here I think the original reading was probably quoipiam, corrupted to quipiam in a, and variously emended by V and C.

§ 94. in eam CE, ineram (with r erased) A, in eram BP'V', in istam P2, in meram V2. Here a must have had eram, CE are no doubt corrected after A.

True text preserved by P alone.

§ 26. conformatum [P], confirmatum others.

§ 66. istuc istac ibit PE (perhaps B'), istuc is tacebit C, istuc ista ibit AVB3, isthaec marg. V same hand. There can be no doubt that P here represents a, and that the form istac being not very common, got corrupted in ABV; the marginal reading of V is probably an attempt to explain istac.

§ 92. hanc PV, haec others.

I go on to give some crucial instances illustrating the character of the different MSS.

§ 13. de Sagra. All the мss have gone wrong here, ACEPV having sacra by inadvertence, but B changes this into sacris to suit the context.

§ 14. commemorabas BPV2, commorabas ACEV'. Here I think ACV represent a corrected in BP.

15. audivi, tibi si ACV, audivit. tu si P, audivi Bis se E.

audivit quam si (quam in ras.) B, The origin of the wrong readings

is misdivision (audivit ibi si), which is corrected with intelligence by B and P, and without sense by E.

§ 16. unus is modus est ACEPV2, for is V' has by a common mistake his, which B alters into ex his to make sense.

§ 18. quoniam esset aliquid in rerum natura...esse aliquid homine melius ACPV, q. esset aliquit hominem aliquid in rerum &c. (si superscr. before esset) B, the eye of the copyist passing from the 1st to the 2nd aliquid, and then si being inserted to give a construction; E has si esset aliquod.

§ 19. ab hac ea questione...separantur ABPV, ea om CE (mistaking it for a superfluous abl. sing.).

$25. aram BCE, aramā A, aranam V', arenam V2, aram aut P. The varieties here can be best explained by supposing a dittographia in a, altered in V and P to get some approach to sense. The correct reading was an easy emendation.

§ 26. aedificatum ACEPV, aedificatum esse B Mu. Sch. The insertion of esse is an easy emendation, which to my mind rather spoils the rhythm of the sentence.

§ 27. naturae ista CEPV, naturae ste A, natura istae B. Probably B is a correction of a, represented by A, ste being carelessly written for sta*.

§ 31. mollis est CEV, mo V', mollest A', molest B', molle est A'B'P. Here a probably had the abbreviated moll-est, written as one word by A, omitted as a monstrosity by V, and altered to the more familiar molest by B.

§ 43. meliora me didicisse...capedunculis...quam rationibus: this is the reading of all the Mss, but C adds refersit to find a government for quam, which it takes for the relative.

Panisci ABPV, Panes CE.

§ 45. sanctissume colunt (with one letter erased before colunt) V, sanctissimi ecolunt A, sanctissimum colunt E, sanctissimae colunt P, sanctissime colunt BC. Here probably V (and a) had sanctissume ecolunt by dittographia of e, this was variously understood by A and P and emended by BCE.

* Deiter (Rh. Mus. 1882 p. 314) gives naturae istae as the true reading of B.

§ 46. fanum est BPV, fanus est ACV' (in A s erased), fannus E. Probably a had fanu st, corrected in BP.

§ 47. lucus AEVB2, lucos P making it the object of interpretor, lucis B' perhaps to suit Athenis, locus C.

$51. Arqui APV', arcui B, arcus AV, arci CE. Arqui is plainly the reading of a, of which the others are corruptions.

§ 56. Argum dicitur interemisse [A2BCE], argentum d. i. A'PV. Probably a had an abbreviation, misread by the first recension (ẞ), but rightly interpreted by B and emended by C.

Aegyptum profugisse [CE], aegyptum profuisse AB'V', aegypto praefuisse B'PV. I think a must have agreed with ABV, and that C and P are emendations.

§ 64. dicamus indigna naturis Madv. conj., dicaliusu ignais ACPV, dic alio usu igneis V, dicamus dignais de dys E., dicamus digna dis B. As m, n, u and li are frequently interchanged in the Mss, there is no difficulty about the first word, but how are we to explain the change from indigna naturis into u ignais? The last syllable is an abbreviation for naturis, so the problem is how to change indignanais into u ignais. The disappearance of the repeated na explains itself and we are left with u followed by hiatus to represent ind. If this is a correct account of the readings in ACPV it is plain that B and E are merely conjectural.

per mare BPV and (with re on erasure of four letters) A, permanere CE and probably A'.

No

§ 65 fin. ni ob rem (see my n.), ni orbem V, niobem AC E, niobe B; in jovem C', an iobem P, om. edd. with some inferior MSS. one has attempted to offer any explanation of the oldest readings; it has been assumed that Niobe in B was a gloss giving the name of the speaker, but why in the world should Niobe have been selected? If ni obrem were the reading of a, this would naturally be changed into the more familiar ni orbem or Niobem, and B would naturally try to make sense by dropping the m.

$69. pestifera est multis admodum paucis salutaris. Instead of est A'BEPV' have sint, CA V sit, in place of the original st; B retaining the plural mends the grammar by reading salutaria.

$ 70.

nocere Deianira voluit ACEPV, n. Dianae ira voluit B by an ingenious emendation.

$ 72.

nec amet BPA V2, necari et A'V'CE. I should suppose that B represents a, misread by ẞ and corrected by P.

§ 74. haec cotidiana, sicae, venena ACEPV, h. c. sica v. B (altering the number to make it agree with cotidiana).

§ 75. (dedisses), dedisse E and before erasure of one letter B.

$79. locum conficit cur di ABEP, locum conficitur di V by carelessness, corrected in V into conficit utrum, in C into conficit ut.

§ 81. septimum consul ABV', septimus c. CE, septies c. PV3, both corrections of a less common construction.

§ 83. qui cum ad by corr. in BV, qui quod ad ABCPV, qui quod cum ad E. The original reading was probably quom.

$86. quasi ego...de fundo...P. Rutilii sim questus ABPV, instead of the last four words CE ingeniously emend protulissem questus.

I add a few examples from the earlier books.

I 5. ut earum C, et earum AE, uetearum B.

etearum with u superscript over 1st letter.

Probably a had

I 17. aecum A (and perhaps a), aequ B, fretum (to govern following judicio) CE.

1 24. ubinam C, ubinan A, ubinon B', ubi BaE. represents a, corrected in B and C.

Probably A

1 36. (vi divina esse), ut divina esse A, ut divinam esse B Oxf., ut divina sit E. It is plain that A is a misreading of a and that B and E are corrections to make some kind of construction.

1 37. (sententia st qui a), sententias qui ABE, sentias qui C (to make a government for the following qui).

1 43. quoius ABE, cujus C.

1 44. fere constat BE, fieri constat ACP. Probably B preserves the original, altered into a more common phrase by the others.

1 49. viderit CE, viderat AP Oxf., videat B. Both C and B

seem to me conjectural emendations of viderat.

1 58. L. Crasso interpolated by BE, omitted by ACP Oxf. and no doubt by a.

I 63. nonne aperte B by corr. from non ea parte, nonne a parte AC, none aperte E. Perhaps in a one line may have ended with none a and the next begun with perte. If so, it would seem that B was corrected from the original.

1 65. doce P, doces ABCE Oxf. I suppose we must assume that P is a correction from doces in a.

1 66. veri tamen similiora A, veri simili tamen similiora PB3 and (reading simile for simili) B'E, veri simile tamen si meliora C. I think P represents a, the scribe of which began to write similiora

out of its place and then corrected it as A.

emendations.

B' and C would then be

1 67. otio A, ocio C, optio BEP Oxf. I think a must have had optio corrected in A and C.

punctis B Oxf., cunctis ACEP.

1 68. ex atomis id natum ACE, ex atomi sit natum B', ex atomis sit natum B2, ex atomis renatum P.

170. dicere turpius BC, diceretur pius P Oxf., diceretur plus AE. Here ẞ, i.e. AP (and V represented by Oxf.), seems to have misread the undivided text of a, rightly read by B and corrected by C.

(nimis callide), nisi callide CE, nisi valde ABP Oxf. I suppose a must have had an abbreviation of callide, misunderstood by the best мss, but rightly rendered by C.

171. quid sit quasi corpus aut quasi sanguis B, quid sit quasi sanguis ACEP Oxf. Another proof that B is descended from a by a distinct line.

1 72. crederem AB E, credem A'B' and probably a, credam P Oxf., credemus C.

olet B, floret ACEP. Forchhammer thinks B an emendation, perhaps a may have had ololet or loet.

1 74. consulto dicis CE, consulta dicis ABP Oxf. Probably C is an emendation.

1 76. nulla alia figura ACEP, nulla in alia f. B by emendation.

1 77.

considera BE, consideras ACP Oxf. and probably a,

B being here an emendation.

1 80. ecquos A, etquos BCE.

182. Apim illum BCE, apud nullum C.

1 95. bipes BCE, impes C.

1 107. Cercopis, Cerconis AE and (by corr. fr. Cratonis) C. B

gives the more familiar cerdonis.

I 114. ne intereat B with a, om. all others.

11 9. peremnia B with a, perennia ACE.

II 10.

crearet B', recrearet AB'CE Oxf. Perhaps there may have been a dittographia of cre in a.

II 11. consules B, quos AC, om. E, vos Oxf. No doubt a had the abbreviated coss.

II 26. liquor aquae declarat C2, l. a. d. effusio APV, l. a. d. effusae B, l. a. d. effusioque E, l. a. dederat effusio C1. Here effusio was probably a gloss on liquor, altered by B and E to make sense.

M. C. III.

d