Only once in the history of gear cutting has there been any record of the cutting of a worm gear weighing ten tons, having 720- 3/4 pitch teeth within an overall tolerance of 0.0001". To three men go the credit for this feat; Mr. G. Sherburne, Superintendent of the Astrophysics Machine Shop at the California Institute of Technology, and his assistants, Mr. Tom Weir and Mr. Lawrence Sills, machinists.

The gear, one of three identical gears cut, is used to drive the mechanism which positions the 200" telescope.

The architect's drawing (Figure 1) shows a cross-section of two of the gears which are employed to position the telescope in right ascension. One of the gears makes it possible to move the telescope quickly in making a setup; the other is the final gear in a set which swings the telescope at the exceedingly slow speed required to follow a star.

The drive gear, which is 14' 3" diameter is of cast iron with 0.25 percent of molybdenum added to give it the necessary 200 Brinnell hardness.

One of the initial problems was the design of a support for the blank which would make it possible to keep the side play down to the error allowance of 0.0001" while the teeth were being cut. The first job, then, was to construct a bearing and spindle that would meet these requirements. This was done by machining the spindle so that it had two tapered surfaces-one near the top, the other near the bottom-that were lapped into two tapered bearings which were made to take them.

A small pilot shaft at the end of the spindle rested on a special thrust bearing which could be adjusted. By

slightly raising the spindle and gear, the coefficient of friction could be lessened to prevent the spindle from freezing in its tapered bearings, although, even with this setup, the matter of lubrication to the spindle caused considerable trouble. Even the lightest oil film would permit side play of more than 0.0001". Indeed, the capillary action of the light oil was sufficient to raise the spindle, gear, and structure-weighing more than 10 tons-as much as 0.0001" The vertical movement was not of much importance, but the side play was. However, by the expedient of machining facets in the tapered sections of the spindle, the error caused by the oil film was eliminated to within the tolerance allowed. Adequate temperature control could be obtained only by the construction of a room about the whole job. Air conditioning equipment was then installed, with which the temperature of the room could be controlled within plus or minus 1° F. When ready to run, the room temperature was maintained at 74° to 76°.

Calculations that had been made showed than any equipment for indexing, if placed upon the gear blank itself, would develop a warping action due to the stresses set up by the cutting, and thus would ruin the accuracy of the job. To eliminate this difficulty, an auxiliary plate of approximately the same diameter as the gear was made and securely bolted to the blank so that measurements could be taken from it for the milling operation.

In order to space the teeth properly, buttons of hardened tool steel, ground to within 0.0001 " of a given size, were mounted on the plate. Dial gages were then used to ensure perfect location.

[Note by Ed., Scientific American.— These "buttons," from here on frequently referred to, are the "toolmakers' buttons" which are familiar to mechanics. Whenever high-precision accuracy in the positioning of a cutting or grinding operation is called for, methods such as making marks by ordinary measuring are not sufficiently close. There must be some additional method of "correcting the aim" by a series of closer and closer approximations. This is done by the use of "toolmakers' buttons." First, the point is determined as closely as possible by ordinary methods. A center-punch mark is made and a hole is drilled and threaded. The button, a small steel cylinder (several of which show in Figures 2 and 4) having a vertical, central hole, is then attached by means of a machine-screw.

The hole in the button is purposely made larger than the machine-screw, so that the button will have side play. If, now, the screw is adjusted just snugly, but not too tight, the button may still be shifted in position within small but sufficient limits. This is done by delicately tapping on its side with a light tool. When the final measurement, made in the present instance by means of a long tram rod, or trammel, indicates that the button is positioned as closely as possible, the screw is set up, and the machine is adjusted to cut in relation to this button, the button itself being then removed to permit completion of the operation. Since the dimension of the button enters into the measurements and calculations, this must of course be accurately known.]

The process of locating the buttons on the button plate presented some interesting problems. The placing of the first one was, of course, easy, but locating the second one, which was to be 180° away, was another matter. All the buttons were mounted on an eccentric, which allowed them to be adjusted slightly. The second button was roughly located with the aid of a straight-edge lined up across the center of the plate to within perhaps 0.004" to 0.005". Then the slow, tedious task of adjusting the buttons at points exactly 180. apart, within 0.0001", began.

Two dial indicators graduated to 0.0001" had already been set up 180° apart, positioned on ground ways. The first indicator was adjusted to the leading edge of the first button, and set to read zero. The second indicator was then set on the second button and set to read zero. Then the gear was revolved 180° and new readings were taken. Of course, the readings were off.

The buttons were then adjusted to compensate for the error, as closely as it could be judged, and the gear was resolved again. The error was less the second time, but still over the allowable limit. Again the buttons were adjusted and again the gear was revolved the 180°. After many more trials and adjustments, the buttons were spaced within the allowable limits.

Now came the problem of locating the second set of buttons at the 90° positions.

The exact spots for these buttons had been determined by calculations and a tram rod had been made to measure the distance from the first two buttons. From the calculations the buttons for the 90° positions were set to within 0.004" or 0.005". Then the tram rod was used.

At one end of the tram rod was a hardened "V," made to fit the buttons. At the other end was a dial gage, reading to 0.0001". The tram was well insulated by wrapping it with paper, and a wooden handle had been incorporated into the construction so that the heat from the machinist's hands would not introduce errors in the readings. Just one degree of temperature shift was found to change the reading as much as 0.0002".

The next move was to adjust the two buttons in the 90° positions so that, by using the tram in any one of the four quadrants of the gear, the error was brought within the allowable limits. This process was repeated with another tram for the 45° positions of the third set of buttons, and other trams of the correct lengths were used until all of the 144 buttons had properly been spaced on the plate. This allowed a button for every fifth tooth on the gear. When the time came for milling the intermediate teeth, an insert of four buttons, shown in Figure 2, was used.

With all the buttons properly located it was time to cut the first gash. Then the gear was revolved 180° to eliminate accumulated error, and the second gash was made. Then to the 90° position, and so on. The rough milling was done in two stages, taking two straight cuts and leaving 0.007" in the second cut to be removed by the hobber in the finishing cut. The hobbing operation is illustrated in Figure 3.

It was found that, during the milling, the heat generated by the cutter materially affected the accuracy of the operation. To relieve this situation, a hollow cutter arbor was used through which water at a controlled temperature could be circulated.

After the first roughing cut had been made for all 720 teeth, the method of checking for accuracy was changed. A microscope was anchored in position over the handwheel of the worm, which was graduated in 0.0001", and a dial indicator was positioned so that it would touch a tooth some distance away on the periphery of the gear as shown in Figure 4. As the gear was always revolved in one direction only, no correction was necessary for backlash. Upon checking and rechecking it was found that the error between any one tooth and any other was only 0.0001".

When the gear blank was made, it had been split horizontally, with the idea that the two sections could be unbolted after the rough milling had been completed, so that the top section could be revolved 180° before the finishing cu1 was taken. This would, of course, have increased the accuracy. However, by using the method and care described above, this action was rendered unnecessary.

The milling operations took ten months and the finish hobbing operation took three weeks. Two and one half

years in all were required to cut the teeth on the three gears. As the work progressed from one gear to the next, the accuracy improved. For instance, in the cutting of the first gear there was an error in one of the quadrants of 0.0004". This error made that particular quadrant unusable However, the three remaining quadrants were within the allowable tolerance and this gear is now used for the North and South declination with the bad quadrant at the top of the gear where it is never used.

This completes the abstract from Robert Clark's account of the job for the 200". In referring to Figure 1 he terms it "the architect's drawing," which is quite correct, but the architect happens to be Russell W. Porter. At that, Porter is an architect-or was so educated.

Those who have created a setting circle-that is, not simply taken off the marks from another divided circle or transferred them from gear-tooth spacings (which is about the same thing), but who have themselves divided a circle out of the void and nothingness, will recognize the almost exact similarity of the method employed with the method described above. The following is a short additional note written for the most recent printing of "A.T.M.A." and inserted on page 302. It briefly explains this method of creating a setting circle, and incidentally may be clipped out and pasted in earlier printings of "A.T.M.A." by those who own that book. It runs-

"Editor's Note: For theoretical data on setting circles, see chapter on 'The Divided Circle,' in Martin's 'Optical Measuring Instruments.' Also see article on 'Divided Circles' in Glazebrook's 'Dictionary of Applied Physics,' Vol. IV. The latter explains the methods used in one example chosen, for inscribing 4320 divisions, one for each 5 minutes of arc, on a 48" circle, within an accuracy of 1/2 second. The method is one of trial and error in a series of closer and closer approximations.

"A crude suggestion of the principle used can be had by tacking a circle of cardboard at its center to a board. Place a reference mark (representing the cutter) on the board adjacent to the edge of the circle and mark the periphery of the circle in continuation of it. Call this zero. Select the 180° point on the circle by approximate estimate. Make a tentative 180° mark, and continue it on the board. Now rotate the circle till the 180° mark lies next to the cutter. The true 180° point should now be halfway between the rotated zero mark on the circle and the nearby mark on the board.

"Estimate new marks for a closer trial and repeat the cycle; and so on.

"Using this as a starter, in the example mentioned above, the 90° and smaller divisions on the 48" circle were filled in by similar principles, and it all required six months of tedious slavery for the 4320 divisions. The approximations soon fell within the field of compound microscopes powerful enough to detect errors of one tenth second, or 1/86,000 inch.

"Gluttons for this general type of masochism will find in the same volume of Glazebrook a 10-page article on 'Diffraction Gratings, the Manufacture and Testing of,' by Dr. J. A. Anderson. Asked how the two jobs, the 48" circle and the construction of a ruling engine, would compare in difficulty, Dr. Anderson stated that the ruling engine job would be far more difficult. since so many more factors enter into it-not alone equality of spacing, but length and parallelism of grating lines."

End of "A.T.M.A." supplementary note.

This is not to say that the circle job is easy, and similarly for the big worm wheel job, but only that the ruling engine job is super, hyper, ultra. The other jobs are merely super, hyper.

EXCHANGE: From Peter Dawson, Care Dr. A. C. Ainsley, Greylands, Victoria Road, West Hartlepool, County Durham, England, comes the following: "Could you find two or three young fellows, at about the same stage as me in telescoptics to correspond and keep the interest aglow, now that the difficulty of obtaining materials over here is increasing. I am only a very green amateur, aged 18, and my interest began three studying physics (with particular accent on optics) and hope to go to University to continue, in October. I have completed a 23/2" refractor, the O.G. of which I did not work-I just did the engineering and woodwork. I have started a 6 1/2" mirror and am at present grinding. I have studied that excellent book 'A.T.M.,' also 'A.T.M.A.' I am also a keen trout fisher. Do you think you could wangle it so that my correspondents are cursed with this plague too?"

LYE will protect a mirror, silvered or aluminized, from sweating due to condensation.-Charles and Harold Lower, of San Diego, report thus: "A warm day followed a night of high humidity, and our mirror sweated. Lye, impure form of caustic soda (NaOH), has an affinity for water, and it is cheap, so we tried putting a can of it inside the telescope when closing up after a spell of observing. It worked. No more condensation on the mirror in the morning. Best of all, we found that it greatly reduced the tendency of the silver to tarnish. We have kept a silvered mirror in good condition for five years by this method, and the telescope was in regular use, in town. About half an inch of lye in the bottom of a peanut-butter jar will protect a 12" telescope for a week or more, in rainy weather, much longer in dry weather. If neglected, it will continue to absorb moisture until the solution is about an inch deep. When the jar accumulates too much water we throw the lye away."

Lye costs about 15 cents a pound, at a grocery store, caustic soda (chemically pure) costs about 15 times that much; and the one is about as good as the other for this. Keep it off the mirror-silver or aluminum-and off the hands.

The Society for Amateur Scientists (SAS) is a nonprofit research and educational organization dedicated to helping people enrich their lives by following their passion to take part in scientific adventures of all kinds.