puts us in touch with a parent document the date of which cannot have been later than the closing months of the year 624, and may have been earlier than 606; for it is uncertain which it was of the three popes bearing the name of Boniface, and living between those two dates, who consecrated the new basilica.

IV. Viewed in connexion with this fact, it, certainly, is a remarkable circumstance that we have no mass in honour of Sancta Maria ad Martyres, a feast instituted by Pope Boniface IV. in the year 610, on occasion of the consecration of the Pantheon.

V. Nor does the Corpus MS. take note of the greater Litanies, instituted by Gregory the Great in the year 598.

Unless, then, we suppose that the monks of St Augustine's, though ready to adopt a multitude of alien feasts, wilfully abrogated two of their most ancient anniversaries, we must allow the terminus ad quem to rest at the year 598. But, even were we to adopt so improbable a theory, we should still be confronted with the Veneratio sancti Michaelis archangeli'; we should still, that is to say, be confronted with a book which left Rome early in the seventh century, which lay concealed nobody knows where till late in the eighth century at the earliest, and which then superseded the authentic Gregorian original, a document known, on the authority of Egbert, to have subsisted, and to have subsisted in several copies, at St Augustine's, Canterbury, from the foundation of that monastery.

Speculations like this verge so closely on absurdity as to bid us beware of them; and I am sure that, on a careful review of all the evidence, my readers will agree with me that it would be an affectation of incredulity to doubt the substantial identity of the parent of the primitive portions of the Corpus MS. with the missals brought to our land in the year 597, and handled by Archbishop Egbert in the course of the eighth century.

THE EXEMPLAR OF THE CORPUS MS.

If, then, as regards those portions of it which are old enough to have been in existence at the close of the sixth century, the Corpus MS. was derived from one of the missals which were brought to England by Augustine and his monks, are we to think it a direct or an indirect transcript?

No one who may have an opportunity of inspecting the Corpus MS. can compare its first few pages with those of its latter half without

perceiving a difference, not in the handwriting, which is the same, but in the writer's manner of wielding the pen. He never in the latter half of the volume seems to be cramped for want of space, and rarely so writes a prayer as to make it fill as few lines as possible, leaving a residuum of words or syllables or, it may be, of one short syllable to be accommodated in the same line with the opening words of the succeeding prayer. But, in the earlier pages this remarkable economy of space, and this remarkable way of effecting it, are of perpetual recurrence; the result being, that in the first few leaves any five lines contain, at the least, as many letters as any six lines in the second half of the volume. I think that if the ruled space in our pages had been half an inch wider than it is, and that if the trammels which bound our transcriber had never been relaxed for the admission of adventitious and intrusive work, this difference would not have come to pass; and the conjecture seems to be a reasonable one, that the penman entered on his task with the intention of crowding a certain amount of text into a given number of lines.

If, then, we suppose him to have been working on narrow columns of uncial character, can it be possible that the task he set himself was that of making a single line of his transcript the equivalent of two lines in his exemplar? The theory is plausible enough; for

I. I. There is no reason in theology or in grammar why he should at fol. 9, lin. 8, have written 'Prope esto domine,' not 'Prope esto'; but the aggregate number of letters in 'A PROPEESTODNE' and the adjacent 'ORATIO' is nineteen, or about half the number of letters contained in a full line of the Corpus writing.

2.

It is hard to see why, but for some such reason as this, he should at fol. 12 v., lin. 7, have stopped where he did, leaving his preposition to govern nothing,-'A. Etenim sederunt principes et aduersum.' These words, with a necessary but omitted OR are of the value of two such. lines as I have indicated.

.

3. At fol. 24, lin. 5, we have the same textual value in 'DOMINICA. II. XL. A REMINISCERE MISE ORATIO.' I cannot otherwise account for this curious truncation of the word 'miserationum.'

4. The theory I have advanced affords the only plausible explanation I can find for the strange 'Uocem jocunditatis annunt,' at fol. 50, lin. 12. These letters with the adjacent rubrics 'DOMINICA · U.' and 'ORATIO' are of the value of two lines of nineteen letters.

II. Pursuing my investigation, I find that our transcriber has two ways of dealing with the syllable or syllables remaining to him over and

above an integral number of lines as he approaches the end of a prayer or preface. Sometimes the residuum is allowed to occupy the beginning of the next line; sometimes it is accommodated at the end, not the beginning, of the next line, the earlier portion of which is reserved for the opening words of a new constituent and for its rubric. Now, when the residuum happens to fall short by ever so little of the full complement of one such line as I have imagined, the transcriber takes the second of these courses (unless, indeed, he be dealing with the last constituent of a mass); but, when it surpasses that complement he takes the first. The theory of such an exemplar as I have imagined affords the simplest conceivable explanation of this remarkable difference.

It will be seen, on comparing my resolution of twenty lines of the Corpus book into the form which I believe their contents to have exhibited in the Canterbury exemplar, that every several detail of title, antiphon, heading and residuary text finds its own proper place with a spontaneity which cannot be fortuitous, which never yields an 'error' of more than a letter or two, and which would have been impossible with lines of any other average content than nineteen letters. I say average content, because the letter I is scarcely equivalent to the half of any other letter, and because, as in the transcript, so, it must be presumed, in the exemplar, the ungrammatical division of a word was on no account whatever tolerated'.

But, I have been anxious to ascertain the number of lines in a page of the exemplar whence the Corpus book was copied, and have no doubt whatever that the number was twenty: for these reasons:

I. 1. At fol. 9 v., lin. 5, our copyist concludes the prayer 'Indignos nos' with 'qui uiuis' instead of 'qui tecum uiuit,' an error which the principal reviser has taken care to correct in the margin. The copyist's blunders, however, are so very rare and, save in this instance, so very slight, that I hesitate to hold him solely responsible for so grave a lapse, and incline to think that the prayer filled the last lines of a page of the exemplar; but (i) that the writer of that book, rather than let 'qui tecum uiuit' travel up into another page, had set it down in the form of some exceedingly compendious abbreviation which the copyist misunderstood; or else (ii) that, though the whole or a part of 'qui tecum uiuit' had once been visible in the original, the formula was by this time obliterated by reason of much handling of the book. The latter is the more plausible alternative, for I find no authority for the

1 My printed lines are not of equal length; nor were those of the original. See M. Ulysse Robert's facsimiles of the Codex Lugdunensis, and his corresponding transliteration.

theory of an exceedingly compendious abbreviation. But, in either case, the most crucial of questions here emerges: Of all possible places for them, can it be that, in providential anticipation of my theory, the words 'qui tecum uiuit,' or a portion of them, fell not only at the extreme end of a page but, as the necessity of the theory requires, at the extreme end of a recto page? I hope to answer this question presently.

2. Resuming the investigation, I find nothing worthy of present mention till we come to the end of the fifth line of fol. 10, where, for ‘qui uiuis,' we have the relative pronoun, indeed, but the relative pronoun followed by 'uiu.' with a horizontal stroke over the third letter. This unparalleled way of writing the word seems to shew that the transcriber knew not what to write; and the view I take of his 'mark of indecision' seems to have been that taken by the reviser, who has been careful, using catchmarks as in the previous case, to write 'qui uiuis' in the adjacent margin. I think, then, that (i) either the words had not been written in full in the exemplar, or else, as before, that (ii) they were not easy of decipherment, and that the reviser's object in making note of them was to communicate to others a piece of knowledge as to which he had no doubt. The latter alternative invites the further inference that another recto page came to an end at this place, and thus at the distance of forty lines of some eighteen or nineteen letters each from the conclusion of the prayer 'Indignos nos.' Hence the theory that the pages of the document which served as exemplar for the Corpus MS. were unicolumnar, and that each page held twenty such lines as I have described. Let us now test the theory.

II. Counting back from the end of 'acceleret qui uiuis' at fol. 10, lin. 5 to the beginning of 'Praesta q. o. d.' at fol. 9 v., lin. 6, we have the transcript of forty such lines; counting back again from the end of fol. 9 v., lin. 5, to the end of 'munere' at fol. 9, lin. 5, another like quantity; counting back once more to the junction of the first and second syllables of 'uenturae' at fol. 8v., lin. 15, the equivalent of twenty. If, then, I am right, the broken phrase 'Praesta qs omp ds ut redemptionis nostrae uen-' filled the last two lines of some multiple of twenty from the beginning of the exemplar. Can this have been the case?

Carefully as the first three masses of the Corpus book have been erased, traces of them remain which enable us to determine how much of them was primitive, and how much adventitious. The first mass contained a Preface in (15-3) 12 lines, the second and third

contained Prefaces in 14 and 8 lines respectively. Their total is 34 lines, which, when deducted from the 75 intervening between the head of fol. 7 and the end of fol. 8 v., lin. 15, leave a remainder of 40 lines. But, from these 40 lines we must make an abatement of 2 lines in respect of the space lost in the ornamentation of the first page. That is to say, the first 40 ruled lines of the Corpus MS. have the value of 38 lines of text. These, in their turn, are the equivalent of 77 such lines as, in my opinion, went to form the exemplar of the Corpus MS.; and, if to that number we add 3 such lines for the value of space lost in ornamentation, we have a total of 80 lines, or four of my hypothetical pages.

I have for my own satisfaction re-cast the first few leaves of our volume into lines such as I have indicated, and, making a column begin with '-tura solennitas et prae-' (fol. 8 v., lin. 16), have grouped them from that point in twenty-line columns, or pages. The third and fourth of these1, representing Page vii. and Page viii. of the original, are:

1 This batch of resolved text begins in the MS. on fol. 9, lin. 16, and ends on fol. 9 v., lin. 15. The collocation on 9 (19) is 'Concede quaesumus omnipotens deus. OR⚫ one consolemur • qui ui-.' The eleventh and following lines of the resolution are illustrated by the accompanying facsimile. The perpendicular strokes in the first ten lines denote the endings of lines in the MS.