TS (a transcription), Mark Twain Papers, The Bancroft Library, University of California, Berkeley (CU-MARK), is copy-text for ‘Westminster . . . I’ (54.1–17); MS, pages 3–4, Harry Ransom Humanities Research Center, University of Texas, Austin (TxU), is copy-text for the remainder (54.17–34); extensive damage has been emended. MS pages numbered ‘3’ and ‘4’, the only ones known to survive, have been silked to prevent further loss, a process that immobilized the several fragments but did not always perfectly align them as they were originally written. A photographic facsimile of the MS is on pp. 452–53. The MS consists of two leaves of off-white wove paper, originally measuring approximately 5¼ by 8⅛ inches (13.2 by 20.8 cm). Inscribed on one side only in black ink (now faded to brown), each leaf has twenty-two horizontal rules. TS is a carbon copy of a modern typescript of the entire letter, including what was presumably on MS pages 1–2, which are lost, but also parts of the text now missing from MS pages 3–4. TS was made in 1940 by Theodore Hornberger while preparing Mark Twain’s Letters to Will Bowen. He typed it not from the MS, which was even then reduced to something very close to its present condition, but from a document he described to Bernard DeVoto as “some kind of a proofsheet or off-print” from the weekly humor journal Texas Siftings, supplied to him “by Mr. R. B. Bowen” (Hornberger to DeVoto, 13 Jan 1941, CU-MARK). When Hornberger published the letter, however, he described this same source as “a typescript of a letter the original of which is incomplete,” and, again, as “a typescript on a letterhead of Texas Siftings, now in the possession of Mr. R. B. Bowen.” Royden Burwell Bowen (born 18 May 1883) was one of William Bowen’s sons; it was he who “presented to the Rare Book Collections at the University of Texas the fragments of the third and fourth pages of this letter, which he found among his papers” (MTLBowen , 3–4, 7 n. 12, 10, 29). Neither Hornberger nor the present editors were able to find the letter in the extant (but broken) file of Texas Siftings, which began publication in 1881 and ceased in 1897, four years after William Bowen’s death. The document owned by Royden Bowen is now lost, and its precise nature (whether proofsheet, offprint, or typed transcription on magazine letterhead) cannot now be verified, even though Hornberger’s published conclusion is probably correct. Collation of MS pages 3–4 with the corresponding part of TS shows that the missing document must have transcribed the letter not only when all four pages were present, but also in a relatively undamaged state. It is clear, for instance, that the references to Miriam and Mrs. Robbins in the final paragraph, which are not now legible in the damaged MS, could not have been supplied solely by conjecture. Since independent evidence confirms their accuracy (see p. 56 n. 7), the only reasonable inference is that the missing document transcribed by TS contained a more complete text derived directly (or indirectly) from the original. Where TS can still be compared with MS, however, it contains several errors in spelling, punctuation, and emphasis, as well as one substantive error, the omission of ‘that’ (54.25). It consistently renders ‘&’ as ‘and’ (ten times) and represents Clemens’s normal em dash as four hyphens (two times). These errors were probably introduced both by the earlier, missing document, and by Hornberger’s typescript of it (TS), for in the transcription that Hornberger published, he reproduced some twelve of them, despite the contrary evidence even of the damaged MS, while he also corrected others, such as all but one ‘and’ for ‘&’, by relying on that same damaged MS. (Hornberger’s handwritten corrections of TS, in ink, are here ignored because they were all drawn from the extant MS, not the missing document of which TS was a copy.) Where TS is the sole source for the text, therefore, it almost certainly contains similar errors in spelling, punctuation, and emphasis which cannot now be confidently identified or corrected. For instance, TS has no underscored words—not even where the extant MS shows Clemens consistently underscoring newspaper names like the World and the Tribune—perhaps because the nineteenth-century typewriter used to copy the letter lacked an underscore key. In other letters written at about this time, however, Clemens omitted the underscore more frequently than he supplied it, and often did both even in a single letter. On the evidence, therefore, unitalicized ‘Tribune’ (54.16) may be accurate, or in error. Only the most likely departures from the lost original are here corrected: ‘and’ is emended to ‘&’ (seven times), and a four-hyphen dash is emended to an em dash (once). On the other hand, where MS is copy-text, but so damaged as to be illegible, TS has independent authority because it derives from the relatively undamaged original. The TS reading is therefore adopted as an emendation, insofar as it is consistent with the remaining textual and physical evidence in MS. All variants between the edited text, MS, and TS are therefore recorded here, even when the reading of TS is manifestly mistaken.