In 1790 William Kelly, manager of New Lanark Mills before Robert Owen came into possession, first applied water-power to the mule, and this at once led to its further enlargement.[389] Taking advantage of the greater driving power available, a Manchester machine-maker named Wright constructed a double mule, which gradually superseded the single mule. With this new construction, which contained about 400 spindles, “the spinner could superintend and operate upon four times the quantity of spindles compared with the former method.”[390] [Footnote 389: _Ibid._, pp. 53-59. Baines, _ibid._, p. 205.] [Footnote 390: Kennedy, _ibid._, pp. 337-338.] The application of water-power did not mean, of course, that afterwards all the operations of the mule were mechanically performed, but, in 1792, Kelly took out a patent for a “self-actor” mule, which he expected young people would be able to operate. In later years the reason he put forward for its not coming permanently into use was that, owing to the introduction of the double mule and the rapid increase in the number of spindles, mule-minding continued to be the task of a man. Apparently there were other reasons, as, notwithstanding numerous efforts, a satisfactory “self-actor” mule was not invented until 1825, when a patent was taken out by Richard Roberts, the famous Manchester machine-maker, who also gave the finishing touches to the power-loom.[391] In the meantime, various other improvements had been effected in the mule, one of which was due to John Kennedy,[392] to whose writings we are indebted for so much of the information we possess of the development of the cotton industry in the later years of the eighteenth and the early years of the nineteenth centuries. [Footnote 391: William Fairbairn in Baines’ _Lancashire and Cheshire_, VI. clxxii. Roberts is an interesting case of a man being the owner of nearly a hundred patents and yet dying in poverty.] [Footnote 392: John Kennedy “was the first to introduce the double speed or twisting motion to Crompton’s mule, and he may be considered as the immediate successor of Arkwright and Crompton” (Fairbairn, _ibid._, cxcvii.).] We have just seen that one consequence of the application of artificial power to the driving of the mule was an increase in its size. Another consequence, closely associated with the one mentioned, was the appearance of the mule in factories, as contrasted with the garrets of cottages, where it had been previously employed. So long as artificial power meant water-power, factories were necessarily erected by the side of streams, mainly in the country districts. When steam-power became available they could just as well be erected in the towns, and with the increasing complexity of machinery the presence there of skilful mechanics, who were lacking in the country districts, was an item of importance.[393]