We have now to explain the operation of the machine by which the thread is formed from the prepared cocoon. At the risk of some repetition, however, it seems necessary to call attention to the character of the work itself. In each prepared cocoon are about a thousand yards of filament ready to pay off, but this filament is nearly as fine as a cobweb and is tapering. The object is to form a thread by laying these filaments side by side in sufficient number to obtain the desired size. For the threads of raw silk used in commerce, the sizes vary, so that while some require but an average of three filaments, the coarsest sizes require twenty-five or thirty. It being necessary keep the thread at as near the same size as possible, the work required is, in effect, to add an additional cocoon filament to the thread which is being wound whenever this latter has tapered down to a given size, or whenever one of the filaments going to form it has become detached. Those familiar with cotton spinning will understand what is meant when it is said that the reeling is effectively a "doubling" operation, but performed with a variable number of ends, so as to compensate for the taper of the filaments. In reeling by hand, as has been said, the size of the silk is judged, as nearly as possible, by a complex mental operation, taking into account the number, size, and state of unwinding of the cocoons. It is impossible to do this mechanically, if for no other reason than this, that the cocoons must be left free to float and roll about in the water in order to give off their ends without breaking, and any mechanical device which touched them would defeat the object of the machine. The only way in which the thread can be mechanically regulated in silk reeling is by some kind of actual measurement performed after the thread has left the cocoons. The conditions are such that no direct measurement of size can be made, even with very delicate and expensive apparatus; but Mr. Serrell discovered that, owing to the great tenacity of the thread in proportion to its size, its almost absolute elastic uniformity, and from the fact that it could be stretched, two or three per cent. without injury, it was possible to measure its size indirectly, but as accurately as could be desired. As this fact is the starting point of an entirely new and important class of machinery, we may explain with considerable detail the method in which this measurement is performed. Bearing in mind that the thread is of uniform quality, it is evident that it will require more force to stretch a coarse thread by a given percentage of its length than it will to stretch one that is finer. Supposing the thread is uniform in quality but varying in size, the force required to stretch it varies directly with the size or sectional area of the thread itself. In the automatic reeling machine this stretch is obtained by causing the thread to take a turn round a pulley of a given winding speed, and