lathe screw, and the other for longer lengths, and in that way there is a tendency to make the wear uniform the whole length of both screws. It will take time to do this for all lengths of lathe screws, but that is what will be required.

Mr. S. W. Robinson.—Mr. President, I have two questions that I would like to have settled in my own mind. Possibly one of them has been settled already in the reading or in the discussion. If so, I did not notice it particularly. One is as to errors occurring in a single turn of the screw-a single thread-a single revolution. Professor Rogers speaks of using a microscope and micrometer screw to correct the position of the cutting tool, as I understand, for errors in the lead screw, as the work goes along. I heard reference once to divisions of a revolution into parts of a revolution— 20ths, and so on. I did not quite catch whether a graduated standard bar was actually used or contemplated in making these screws, the graduation of which was fine enough, say, to divide the pitch of the screw being cut into 20 parts, and then stop at each graduation mark, or not to stop, either, but watch for the 20ths of the revolution and adjust the position of the tool at each mark. There might be a signal, for instance, giving notice of the approach of each 20th of a revolution. Every screw revolves so slowly in the cutting that there is time for a signal to be made and the observer to watch his opportunity and adjust the thread in each 20th of a revolution, if you please, or in any other number of divisions. It is evident that it can be carried to that extent, and I am not certain whether it was intended to be explained as if carried to that

extent.

Now, this subject touches, perhaps, a point of interest in regard to the permanence of metal; that is, with respect to the bar, the permanence of the bar itself. If we could easily obtain a bar which could be divided into 20ths of a thread for its whole length and be perfect, it strikes me that this method of cutting a screw would result in cutting a perfect screw, that is, perfect within our present means of measuring, or present notion of precision in such matters, to say the least.

As to the permanence of metals, that question, if it is touched upon, seems to be an important one here, not only to the standard bar, but to the screw which is to serve as a standard. A screw that has been made with great care, of course, should be permanent as to its constitution and dimensions. Dimensions are the leading point. If there is a change in the physical constitution in any way,

such as in the relation of the molecules, there would be a change in the dimensions of the screw. I was once struck with a remark in

a letter I received from a young man employed at the Elgin Watch Works. He said that there, in making mandrels that they desired to use in work of great precision, they were in the habit of getting out the blocks of steel nearly to size, tempering them, if they were to be tempered, then laying them away a year or two to season. Now, it seems to me that if there is anything in that, if it is necessary to season steel, it is quite necessary that these standard steel screws should be "seasoned," and also the standard graduated bars to be used in guiding the screw cutting, whether it be carried to the extent of divisions of the revolution or not. I would like to ascertain what has been done in regard to those two points.

Prof. Rogers.-The 6 foot screw in the new machine which has been referred to was cut in the way which Professor Robinson has described. It was not quite feasible to employ the principle of reversing the half nuts in a screw having this length. Since the provisional leading screw was cut five threads to the inch and the bar was graduated to tenths of inches, it is obvious that the periodic errors depending on half revolutions of the leading screw were eliminated, but this operation is rather tedious and requires a great deal of care. Hence, in the new machine, there is an automatic device for correcting the periodic errors which depend upon single revolutions of the screw. The larger share of the errors of this class are eliminated by means of a template. For the elimination of those which remain, dependence must be placed upon the principle of the reversal of a nut divided into sections. Professor Rowland has not yet published a description of his process of correcting a screw, but it is understood that he divides the nut into four sections. A short nut of this kind has recently been made for me by Mr. Ballou, and its action is very satisfactory. I suppose that Professor Rowland has anticipated me in the use of a nut having four sections. I may or may not have anticipated him in the use of one having two sections. When these sectional nuts are placed in position upon the screw, the faces of the threads of the screw and of the nut are everywhere nearly parallel. They may even touch at every point, but if there are periodic errors in the screw, the relation between the threads of the nut and of the screw will vary in every part of the revolution. By repeated reversals of the sectional nuts, therefore, the tendency will be to equalize the errors at every point.

Mr. Kent.-There has been a suggestion made about moving the leading screw along the lathe from end to end, and letting it project beyond the ends three or four feet, in order to have the nut move at all parts of the leading screw. I have a suggestion to make as to how that might be accomplished without moving the leading screw. Let the leading screw revolve in its own place all the time, and have a secondary bar or shaft above it, which may project beyond the end of the lathe, to which the tool post is attached. The nut of the leading screw is to be clamped to this shaft in any part of its length, and the position of the nut on the screw changed from time to time so as to wear the screw uniformly. The shaft slides between bearings so that it shall always pull in a straight line (illustrated on the blackboard).

Mr. Porter.—I would like to make one inquiry of Professor Rogers, and that is whether any attempt has ever been made to remove the periodic errors by a system of scraping instead of grinding. The reason I ask that question is this: It is obvious that when a perfect screw has been produced it is desirable that it be maintained as long as possible, that it shall have as long a lifetime and be capable of as much service in producing other perfect screws as possible. Now, any grinding material whatever finding its way into the metal makes it a permanent lap. If that can be avoided, it seems to me that a longer lifetime would be insured to the screw, and it has seemed to me, without having gone into the subject in my mind very carefully, that a system might be devised by which high points might be detected upon the face of the screw in the same manner as they are detected upon a plane surface and removed by scraping, which can be done in a manner exceedingly delicate. My inquiry is whether any attention has been given to that-whether any effort has been made in that direction.

Prof. Rogers.-Yes, the experiment was tried by Mr. Van Woerd, but without success. I do not think the plan would be possible. We do not deal with plane surfaces in cutting a screw. We deal with helices.

The objection to the use of emery in grinding is not a serious one in practice, especially if the steel is hard. We shall yet have hardened steel screws. Tempered steel behaves quite as well as annealed steel under variations of temperature. I regard the coefficient of my hardened steel standard as more securely determined than for any other metal except perhaps for copper.

Mr. Bond.-It may be hardened outside, and not in the center.

That is possibly the effect of a change of the co-efficient of expansion, because the main body of the bar is in a normal condition, and probably not very hard. We know that in standard cylindrical ganges that are made of tool steel, there are changes that come about that seriously affect the diameter of the ring which fits the gauge, but it is only in steel that is of very high carbon.

Prof. Rogers.-I shall still be obliged to say that I do not believe that the action which Mr. Bond describes is the result of a real change. All of my observations point to an absolute constancy in length soon after the steel is tempered. It cannot be possible that the effect of a change of two or three degrees in temperature at the surface of a steel plug two inches in diameter, will be to change the absolute diameter within a few seconds of time. Is it not more reasonable to suppose that the apparent increase of diameter may be due to the development of surface heat through surface friction in inserting the plug into the ring?

The principle of conservation of forces applies here as elsewhere. A change in the entire mass requires a certain amount of time, but there is some evidence that strictly surface changes may occur, and it is possible that the observations made by Mr. Bond may be explained in this way. The experience of Alvan Clark & Sons in applying the principle of local corrections in finishing an objectglass gives some color to this theory of local surface changes. Even in an object-glass of the largest diameter, two or three polishing strokes with the finger will change the curvature at that point only, by an amount which can be instantly detected by the eye in the optical tests employed. Are we to say that the slight change in temperature produced in the operation of polishing has affected the entire mass of glass? We might more reasonably go to the other extreme and say that in the operation of polishing the direction of the reflecting particles of glass which compose the surface is changed, and that the actual change in position is therefore enormously magnified by the change in the direction of the rays of light which reach the eye. It must, however, be clearly understood that this notion of surface changes, which are distinct from changes which occur in the whole mass, is only a tentative explanation of a very difficult problem. These slight changes are strictly local, and cannot in any sense be considered as affecting what we call the permanent "set" of metals. The observations of Mr. E. S. Wheeler show that in the case of zine there may be a change of "set," but not in glass or steel. But in these experiments the change of tem

perature was very violent. The bar of zinc after having been packed in melting ice was quickly removed and plunged into boiling water. After remaining in boiling water for two or three hours the bar was re-packed in ice. Under the changes of temperature which occur in ordinary experience, the zinc bar should assume its normal length in melting ice within 15 or 20 minutes, but Mr. Wheeler found that after the bar had remained in the ice bath for 24 hours, it was still much too long, and it was only after an immersion of three days that it assumed its normal length. These experiments were very carefully made, and probably a repetition of them would give nearly the same results which Mr. Wheeler found. A small part of the amount of change of "set" may possibly be due to the fact that the bars were not wholly immersed in the bath. It is my experience that the packing of ice should cover the bar to the depth of three or four inches, but the error which may arise from shallow packing cannot much exceed five or six mikrons, or about inch.

As far as I can learn, the positive evidence of "set" in any published record, is limited to the paper by Mr. Wheeler, and here the evidence is limited to a composite metal, viz. zinc.

Mr. Oberlin Smith.-Mr. President, I want to suggest a possible method that has just occurred to me-perhaps, though, it will all be knocked into a cocked hat by the experience of some of the gentlemen-of making a prime leading screw from which to cut other screws. Suppose we had a large cast-iron drum, D, Fig. 65,-so

large that the flexure would not amount to anything. It could be made hollow for strength and to get less dead weight. Suppose it

is a foot in diameter, and four or five feet long. Suppose that into that, in a spiral line as nearly as could be drawn, there were set a