In a subsequent and stronger machine, the bar carrying the mandrel stood lower than the other, to admit of larger change wheels upon it, and the same driving gear was retained. And in another structure of the screw-cutting lathe, Mr. Maudslay placed the triangular bar for the lathe heads in the center, whilst a large and wide slide-plate, moving between chamfer bars attached to the framing, carried the sliding rest for the tool: in this last machine, the mandrel was driven by steam power, and the retrograde motion had about double the velocity of that used in cutting the screw. Indeed these machines may be fairly considered to be the precursors of the present screwing lathes, in which the detached triangular bars or slides have been exchanged for one strong bearer with two ridges or fillets, upon which the slide plate moves for guiding the traverse of the tool.

The relations between the guide-screw and the copy were varied in all possible ways: the guide was changed end for end, or different parts of it were successively used; sometimes also two guide-screws were yoked together with three equal wheels, their nuts being connected by a bar jointed to each, and the center of this link, (whose motion thus became the mean of that of the guides,) was made to traverse the tool. Steel screws were also cut and converted into original taps, from which dies were made, to be themselves used in correcting the minor errors, and render the screws in all respects as equable as possible. In fact every scheme that he could devise, which appeared likely to benefit the result, was carefully tried, in order to perfect to the utmost, the helical character and equality of subdivision of the screw. Mr. Maudslay succeeded by these means, after great perse-verance, in making a very excellent brass screw about seven feet long, and which, compared with standard measure, was less than one sixteenth of an inch false of its nominal length. Taking the error as the one-thousandth part of the total length of the screw, which was beyond its real quantity, to make from it a corrected screw by the system of change wheels, would have required one wheel of 1000 teeth, and another of one tooth less, or 999; but in reality the error was much less, and perhaps nearer the two-thousandth of an inch; then the wheels of 2000 and 1999 teeth would have been required; consequently the system of change wheels is scarcely applicable to the correction of very minute errors of length.

The change of the thousandth part of the total length, was therefore given to the tool as a supplementary motion, which might be added to, or subtracted from the total traverse of the tool, in the mode explained by the diagram, fig. 612, in which all details of construction are purposely omitted. The copy C, and the guide-screw G, are supposed to be connected by equal wheels in the usual manner; the guide-screw carries the axis of the bent lever, whose arms are as 10 to 1, and which moves in a horizontal plane; the short arm carries the tool, the long arm is jointed to a saddle which slides upon a triangular bar i i.

Fig. 612.

In point of fact, the tool was mounted upon the upper of two longitudinal and parallel slides, which were collectively traversed by the guide-screw G. In the lower slide was fixed the axis or fulcrum of the bent lever, the short arm of which was connected by a link with the upper slide, so that the compensating motion was given to the upper slide relatively to the lower.

The triangular bar i i, when placed exactly parallel with the path of the tool would produce no movement on the same, and C, and G, would be exactly alike; but if i i, were placed out of parallelism one inch in the whole length, the tool, during its traverse to the left by the guide-screw G, would be moved to the right by the shifting of the bent lever one-tenth of the displacement of the bar, or one-tenth of an inch.

Therefore whilst the guide-screw G, from being coarser than required, moved the principal slide the one-thousandth part of the total length in excess; the bent lever and inclined straight bar i i, polled back the upper or compensating slide, the one-thousandth part, or the quantity in excess; making the absolute traverse of the tool exactly seven feet, or the length required for the new screw C, instead of seven feet and one-sixteenth of an inch the length of G. To have lengthened the traverse of the tool, the bar i i, must have been inclined the reverse way; in other words, the path of the tool is in the diagram the difference of the two motions; in the reverse inclination, its path would be the sum of the two motions, and i i being a straight line, the correction would be evenly distributed at every part of the length.*

Whilst Mr. Maudslay s experiments in perfecting the screw were being carried on, his friend Mr. Barton,† paid frequent visits to his manufactory, and also pursued a similar course. Mr. Barton preferred, however, the method of the chain or flexible band, for traversing the tool the exact quantity; because the reduction of the diameter of the pulley or drum, afforded a very ready means of adjustment for total length; and all the wheels of the mechanism being individually as perfect as they could be made, a near approach to general perfection was naturally anticipated on the first trial. This mode, however, is subject to the error introduced by the elasticity or elongation of the chain or band, and which is at the maximum when the greatest length of chain is uncoiled from the barrel.

These two individuals having therefore arrived, by different methods, as near to perfection as they were then respectively capable of; each made a screw of the same pitch, and 15 inches long, and the two when placed side by side were found exactly to agree throughout their length, and were considered perfect. The two screws were submitted in 1810 to the scrutiny of that celebrated mathematical instrument maker, the late Mr. Edw. Troughton, F.R.S., etc, who examined them by means of two powerful microscopes with cross wires, such as are used for reading off the graduations of astronomical instruments; applied like a pair of the most refined compasses, to measure the equality of some 20, 50, or 100 threads, taken indiscriminately at different parts of the length of the screws.* From this severe trial it resulted, that these screws, which to the unassisted sight, and for almost every purpose of mechanism, were unexceptionable, were found to be full of all kinds of errors, being unequally coarse at different parts, and even irregular in their angles, or "drunk." This rigid scrutiny led both parties to fresh and ultimately successful efforts, but of these our limits will only allow us to notice one, apparently derived from the use of the two microscopes.