About the commencement of the present century, the turning lathe received material alterations and amendments from many different hands. Among them, the headstocks and the square iron mandrel hitherto in almost exclusive use, were remodelled by the late J. J. Holtzapffel, 1794, who commenced by constructing the old wood formation of the lathe head in brass, a material at that time more easily worked than iron. His earliest structural improvement consisted in connecting in one the two independent pieces forming the headstock, which gave the base and the two uprights the character and advantages of three sides of a square. This rigidly maintained the coincidence given to the bearings or collars in their first construction, and produced so effective a reduction in lateral wear in the collar, that the improved form of headstock soon became nearly universal; while some of his early lathes may still be met with in satisfactory use. Subsequently, the advantages offered by cast iron and the comparative facihty acquired in working in this metal, induced the same engineer to attempt the reformation of many other portions of the old wooden machine; which, as they became constructed in metal, were also divested of their rectangular outline hitherto used, and indicative of the carpenter's plane and the age of wood. The importance of the changes in the external form of the lathe however, may be considered as slight, in comparison with the alterations in the form and material of the mandrel, collar and other working portions of the mechanism, which also were all gradually made of steel.
The earlier iron mandrel running in an iron collar plate, fig. 65, afforded facilities for construction, but these were more than counterbalanced by its disadvantages. The wear between the cylindrical neck of the mandrel and the collar, arising from the section and from the unfavorable nature of the material, being also to some extent accelerated by the square section of the shaft. The work being firmly attached to the end of any lathe mandrel, the two may be considered to form one solid, and when the tool is cutting, to constitute a bent lever of which the mandrel collar is the fulcrum; but, as the other extremity of the mandrel is fixed and cannot escape laterally, the force or resistance to the tool is all exerted as if to bend the mandrel about the middle of its length. The square mandrel is found to yield differently in the resistance that it offers to the cut of the tool, as the revolution alternately presents its diagonal and its square section, and in some cases the results may be visibly traced upon the work; while the variation tends to create irregularity in, and also to accelerate the ordinary wear that takes place between the mandrel and the collar.
Wear between any collar and mandrel causes the latter to float about at random laterally, to the extent of the difference of diameter that has arisen between the two, and this interference with the truth of revolution, is styled "play" in the collar. The disagreement will become in time both audible and apparent, and the mandrel instead of running silently then vibrates under the tool, making a low growling noise at small velocities, and a shrill rattle at high speeds; the vibrations of a much worn or ill fitting mandrel, also sometimes filling the surface of the work with waved lines or striae.
The mandrel of circular section fig. 79, replaced the square, and as this section is always constant, such a mandrel exhibits no variation during revolution, and does not bend except from weakness; that is, when its proportional area is too small for its length. Wear was further reduced by making the collar and the mandrel of steel, and then still more by alteration from the cylindrical to a conical fitting; this latter, considerably delays disagreement in diameter, as the mandrel may be further advanced into the collar by the end screw to compensate the wear, as that takes place.
The cylindrical collar however, had the advantage of admitting the largest diameter possible for the screw for the chucks, and for the face of the mandrel against which they bear when screwed in their place, and this it was very desirable to retain. A mandrel made with an acute cone fig. 80, would allow the screw or "nose," and the face, to remain nearly as large as before; but this form is inadmissible, as it would wedge or set fast in the collar on revolution from the acuteness of the angle. A more obtuse cone fig. 81, could not set fast, but it would so far reduce the face of the mandrel and the nose, as to render the latter, weak and disproportionate to the dimensions of the former.
The form of bearing for the mandrel fig. 82, has two cones in juxtaposition, which avoids the inconveniences and combines the advantages indicated by these diagrams. The front cone is acute, differing but little from the cylinder, allowing a large diameter for the screw and a sufficient shoulder or face for the chucks; the second cone is obtuse, of about the angle of 45°, and narrow in width, its bearing preventing the acute cone from wedging. The value of the narrow obtuse cone is very observable in this respect in the course of construction, when in fitting the mandrel the acute cone continually sets so fast in the collar, as to require a blow from a lead or tin hammer on the end of the screw to drive it back; but this effect instantly ceases so soon as the obtuse cone takes its bearing. The angles and proportions of the back center mandrel and collar indicated by fig. 82, were adopted as the most successful result of an extended series of trials, made by the same engineer and his son, the late Chas. Holtzapffel; they have since been very generally employed, while experience has shown their complete efficiency for lathe mandrels of moderate dimensions.