532 APPENDIX B wrote in 1724, and they contain the names of those knights only who were actually installed, and appear to omit also those knights who were sub- sequently removed from the Order through attainder or for other reasons. The many gaps which consequently arise have to be filled by more or less felicitous deduction from inadequate premisses, much assistance being derived from comparison of successive Great Wardrobe accounts, which specify in several years the names of the knights to whom Garters were sent for the annual feast on St. George's Day. The inherent difficulty in the task of framing a satisfactory order of succession is not diminished by the custom in early times of trans- ferring knights from one stall to another, which is responsible for some- what chaotic results obtained from studying carefully the Windsor Tables and the arrangements of the stalls to be found at various dates in con- temporary MSS., many of which are printed by Anstis. The complete re- conciling of such results seems to be quite beyond the wit of man to accomplish. From the time of Henry VII, however, the custom has been (except in the cases of Princes of the Blood and foreigners admitted to the Order, who for more than a century have been regarded as Supernumerary and therefore negligible in this connexion) to assign each new knight to the lowest stall at his election, and from the reign of Charles I (") (and probably earlier) the knights who were subjects of the Sovereign and not of royal blood ranked in order of seniority of election. Beltz devoted much industry and careful research to the task of fixing the exact succession of knights, and he achieved so great a measure of success that his arrangement has hitherto held the field and has been accepted without question, and even Dr. Shaw, though most pains- taking in his researches, has been too modest to make any amend- ments in it. An examination of Beltz's list shows that his plan was, save in some exceptional cases, to describe each knight as the successor of that predecessor whose place, if there was more than one vacancy, had been longest unfilled. This course, which is in accordance with common sense, has been adopted in the following pages. Beltz, however, — being, like other writers, not infallible — made some considerable errors in the dates of death of certain knights, and Dr. Shaw, though he discovered and corrected these, unfortunately failed to observe the effect which these errors necess- arily had upon Beltz's allocation of those knights' successors. The most striking errors are in the cases of the Earl of Mar (No. 397) and the Earl of Kellie (No. 410) to whom he assigned 1654 and 1643/4 as dates of their respective deaths, whereas he should have given 1634 for the former and 1639 for the latter as Dr. Shaw corrects. C') (^) See Heylin's History of St. George, 2nd ed. 1633, p. 402, where the Knights of the Order on 20 Oct. 1632 who were natural-born subjects of the Sovereign are given in order of seniority of election. C') In each of the following cases Beltz, whose order of succession Dr. Shaw adopts throughout, has given erroneous dates of death: these, except in the case of Prince Maurice (443), have been corrected by Dr. Shaw, who has, however, unluckily
