1 An apology or at least an explanation is due with regard to the manuscript notation adopted in the present article. This is itself but a fragment of a more extended inquiry, and for this I found the current symbols unsatisfactory. There is, indeed, no fixed and consistent usage, and those generally employed are both clumsy and confusing. The notation I propose and have here employed is as follows: every manuscript in the British Museum is designated by a single capital letter (the initial, as a rule, of the collection in which it is found); every manuscript at Oxford or Cambridge by a capital and a small letter (the Oxford manuscripts having the two letters different, the Cambridge the same); every other manuscript by a Latin capital followed by a small Greek letter; where there are two or more manuscripts in the same collection they are distinguished by superior numerals; prints are indicated by the initial of the printer followed by the decade number of the date. I have not yet examined all the accessible manuscripts containing the beginning of the Knight's Tale, still less all extant, but have made a beginning with those in the British Museum, a few at Cambridge, and all those that have been facsimiled or reprinted. There seem to be 68 manuscripts of the Canterbury Tales (or parts of them) known, of which 57 are believed to contain a portion at least of the selected passage. Of these I have collated 22; the following list giving their identifications and symbols:

A¹ = B.M., Addit. 5140

A³ = B.M., Addit. 35286

E¹ = B.M., Egerton 2726

E² = B.M., Egerton 2863

E³ = B.M., Egerton 2864

H¹ = B.M., Harley 1239

H³ = B.M., Harley 1758

H⁶ = B.M., Harley 7333

H⁷ = B.M., Harley 7334

L = B.M., Lansd. 851

R¹ = B.M., Royal 17 D. xv

R² = B.M., Royal 18 C. II

S¹ = B.M., Sloane 1685

S² = B.M., Sloane 1686

Cp = Oxford, Corpus 196

Dd = U.L.C., Dd. 4. 24

Gg = U.L.C., Gg. 4. 27

Tt¹ = Camb., Trinity R. 3. 3.

Tt² = Camb., Trinity R. 3. 15.

Eλ = Ellesmere

Hγ = Hengwrt 154

Pτ = Petworth

Whenever in my article I speak of ‘no manuscript’ or ‘all manuscripts’ I do so strictly with reference to this list; readers will kindly understand ‘manuscripts consulted’ not ‘manuscripts known’. Whether any further installments of the inquiry I have undertaken are published must depend mainly upon whether any results of value seem likely to emerge, a possibility which at the moment appears to me rather remote. Meanwhile, however, I think that some interest attaches to the solution of the more restricted problem.

2 Specimens of all the Accessible Unprinted Manuscripts of the Canterbury Tales, 1890, &c.

3 The table requires some explanation. Although mere variations of spelling are excluded, whenever a variant is quoted the forms found in the different texts are reproduced exactly. Where no reading is entered exact agreement with that next preceding is to be understood. For the sake of clearness, however, substantial variants are distinguished by heavy type, the rest being mere differences of form. Such a distinction is of necessity arbitrary, but is useful in practice. Here the absence of a final ‘e’ required by the metre (but not of one required by grammar) had been treated as a variant, otherwise its presence or absence has been held indifferent. Word division and capitalization has been similarly disregarded. A caret-mark indicates the omission of a word. A letter or word in parenthesis serves to indicate the position of a reading in the line. Letters printed in italic are represented in the original by some mark of contraction. The tags found on certain final letters in the manuscripts, originally indicating an ‘e’, have been disregarded, since they have evidently degenerated into mere scribal tricks. Those found in Caxton's prints, particularly with the letters ‘d’ and ‘h’, have also been disregarded, but an exception has been made in favour of that distinguishing final ‘g’, which may possibly be significant and has been rendered by ‘e‘. For the Corpus manuscript I have relied on the Chaucer Society's print. To facilitate reference, where several variants appear in the same line, they have been distinguished by superior numbers.

4 I have not examined the manuscript at New College, Oxford, (Nw) which is said to be the closest to Tt², but Zupitza is quite clear that it is with Tt² and not Nw that C₇ most closely agrees.

5 It is to be presumed that some of the four readings peculiar to C₇ are to be found in Nw, but the greatest possible number of exclusive correspondences afforded by that manuscript is four. No doubt some of the exclusive correspondences of C₇ and Tt² are also to be found in Nw, but it seems exceedingly unlikely that they would amount to as many as sixteen.

6 It might have been a daughter of Tt² in which a certain number of readings had been introduced from another source, say Cp. But if we allow the assumption of conflation in hypothetical manuscripts a definite conclusion becomes impossible. It is only through the exclusion of such a possibility that the statement that B is (or is not) descended from A can ever be significant.

7 It will be best to count 70^1,2, 73^1,2' and 73^3,4 each as a single reading, thus reducing the number to twenty-three.

8 The other thirty-four are 2², 13¹, 25², 26, 31², 48³, 55², 56¹, 62¹, 65¹, 68¹, 71, 75¹, 76², 85, 87², 87², 89², 91⁴, 92, 93⁴, 94⁴, 95¹, 95², 96¹, 96², 97¹, 97², 98², 99, 100¹, 101, 111, 114¹.

9 Or C₇, though in that case we should expect to find some distinctive C⁷ readings, which we do not. So far as the eight significant readings are concerned the restorations might equally well have been made by comparison with W₉, and this would seem on the face of it more likely. The supposition, however, is ruled out by the fact that P₂ occasionally restores the archaic ‘hem’ for the ‘them’ of P₉ in cases (87³, 95³) where W₉ modernizes. Since we can hardly suppose that the compositor of P₉ worked with a manuscript as well as C₈ in front of him, we must suppose that his copy of C₈ had already had manuscript alterations inserted. If so it was presumably not the same copy of C₈ that was used in the preparation of P₂.

10 The Chaucer Society print of the Petworth manuscript is ambiguous. In this line the word ‘ther’ is enclosed in brackets, but what is meant thereby is not stated.