The reel is now put into movement, and winds the thread formed by the union of the filaments. It is at this moment that the real difficulties of the reeler begin. She has now to maintain the size and regularity of the thread as nearly as possible by adding new filaments at the proper moment. The operation of adding an end of a filament consists of throwing it in a peculiar manner on the other filaments already being reeled, so that it sticks to them, and is carried up with them. We may mention here that this process of silk reeling can be seen in operation at the Manchester exhibition. It is only after a long apprenticeship that a reeler succeeds in throwing the end properly. The thread produced by the several filaments is itself so fine that its size cannot readily be judged by the eye, and the speed with which it is being wound renders this even more difficult. But, in order to have an idea of the size, the reeler watches the cocoons as they unwind, counts them, and, on the hypothesis that the filament of one cocoon is of the same diameter as that of another, gets an approximate idea of the size of the thread that she is reeling. But this hypothesis is not exact, and the filament being largest at the end which is first unwound, and tapering throughout its whole length, the result is that the reeler has not only to keep going a certain number of cocoons, but also to appreciate how much has been unwound from each. If the cocoons are but slightly unwound, there must be fewer than if a certain quantity of silk has been unwound from them. Consequently their number must be constantly varying in accordance with their condition. These facts show that the difficulty of maintaining regularity in a thread is very great. Nevertheless, this regularity is one of the principal factors of the value of a thread of "grege," and this to such an extent that badly reeled silks are sold at from twenty to twenty-five francs a kilogramme less than those which are satisfactorily regular.