of the abbey of St. Austin at Canterbury. To these I may add accounts in Registrum Roffense by Mr. Thorpe, other pages of registers not published by him, and the consistorial and archidiaconal acts of that diocese; and I am much mistaken if in any of the accounts a coeval Arabic figure will be discovered before the conclusion of the sixteenth century.
Mr. Boys, who has had access to all the records of Sandwich, and who, while he was writing his history of that town, and of many places connected with it, examined the books and papers with the closest attention, acquaints me that no Arabic characters occur in the accounts before the middle of the sixteenth century; and in the old churchwardens accounts of the parish of Lambeth, which begin in the year 1505, there is not for several years a common figure inserted, all the entries being in Roman characters. The churchwardens accounts of Shome, a parish not far from Rochester, are entered in Roman numerals as late as the year 1621; and yet it is observable, that the common figures are used to mark the date of the year 1556, the second and third years of the reynes of Philipp and Mary, kyng and quene. For private accounts I may refer to the Northumberland House Book, wherein the same numerals are used; and in the original letters published by the late sir John Fenn, "the respective sums which William Ebesham had received for transcribing books for his worschipful master sir John Paston, and what was paid about the year 1459[1]," are in the same characters. And should it be remarked that these are only negative proofs, the answer is obvious, that such a series of them as might be produced will preponderate, whilst there is not, as yet, any positive proof to put in the opposite scale.
But as the chief evidences in support of an early introduction of Arabic numerals were drawn from inscriptions, some rudely executed, others mutilated, and others where the supposed figures al-
- ↑ Vol. II. p. 10, et seq.
luded
