[74] Ibid., Vol. I, pp. 35-6.
[75] Familiar Letters, Vol. I, p. 244. See also Lockhart, Vol. V, p. 408.
[76] Sometime before 1821 (probably a good while before, but the date cannot be fixed), Scott began a translation of Don Quixote, and afterwards gave the work over to Lockhart, who completed it. See Constable's Correspondence, Vol. III, p. 161.
[77] Louis-Elizabeth de la Vergne, Comte de Tressan, was born in 1705 and died in 1783. In early life he was sent to Rome on diplomatic business, and it is said that in the Vatican library he acquired his taste for the literature of chivalry. His chief works were Amadis de Gaules (1779); Roland furieux (translated from the Italian, 1780); Corps d'extraits romans de chevalerie (1782). His translations were partly adaptations, and were far from being rendered with precision.
[78] See particularly his article on Ellis's and Ritson's Metrical Romances (Edinburgh Review, January, 1806), the essay on Romance, and Remarks on Popular Poetry in the Minstrelsy.
[79] Edinburgh Review, July, 1804. Ellis and Scott had had much correspondence on Sir Tristrem, and it was Ellis's queries that first led Scott into the detailed investigation which resulted in the separate publication of the work. He had intended to print it in the Minstrelsy (Lockhart, Vol. I. p. 289). The letters are given in Lockhart, Vol. I.
[80] Lockhart, Vol. I, p. 381.
[81] Die nordische und die englische Version der Tristan-sage—II. Sir Tristrem. Heilbronn, 1882. Mr. George P. McNeill's edition of Sir Tristrem was printed for the Scottish Text Society, Edinburgh, 1886.
[82] Kölbing thinks Scott probably hired a transcriber who knew nothing of Middle English—a usual method of procedure in the beginning of the nineteenth century. In later editions more errors were introduced by the carelessness of printers, until, after 1830, when the book was included in the complete editions of Scott's poems, the text was collated with the manuscript. But it was still far from correct. Kölbing enumerates about a hundred and thirty mistakes (see his Introduction, p. xvii). Of these I took twenty-one at random, and found that eight of them did not occur in the 1806 edition—in other words, the person who collated the text nearly thirty years after Scott or his hired transcriber had done it was far from infallible. A few illustrations may be given of mistakes that occur in both the 1806 and the 1833 editions: l. 117, send is given for sent; l. 846, telle for tel; l. 863, How for Hou; l. 912, mak for make; l. 1212, leuedi for leuedy; l. 1580, wende sche weren for whende sche were; l. 1334. have for han; l. 1514, as for als.
[83] Review of Johnes's Translation of Froissart, Edinburgh Review, January, 1805.