The general plan for the large aerodrome was never a matter of uncertainty. At the time when the first general designs were made there had been in the history of mankind only one type of machine, that of the steam-driven Langley models, which had proved capable of flight for any considerable distance. Furthermore, the selection of this type had been the result not of sudden fancy or of purely theoretical consideration, but of years of the most careful experimentation, in the course of which nearly every conceivable style of machine had been tested with some form of power. It would have been worse than folly, therefore, if the one clear path had been left to seek some unknown way.

It was fully realized from the first, however, that the increase in size alone would make necessary in the design for the large aerodrome a great many modifications from the designs of the steam-driven models. It was not possible here, as in nearly every other kind of structure, simply to magnify uniformly the parts and proportions of the small machine in order to obtain a successful large one. This is particularly true in the case of the aerodrome, because the rapid increase of weight in the larger structure is out of all proportion to the increase in strength, while it is very desirable that the more expensive machine which is designed to carry a human being shall be relatively even stronger than the easily replaced model. This problem of increasing size without sacrificing strength and stability, it was known from the beginning, would be encountered in a particularly difficult form in designing the frame of the large machine, and was to be solved not by the discovery of some new and wonderfully strong material, but by improvements both in the general plan and the details of the machine. Here, as is often the case, it was not the large changes in the design but the improvements in small and sometimes seemingly unimportant details which demanded the most careful consideration and, as a whole, contributed most to the final result. For this reason, as well as because the large changes, when pointed out, are usually easily understood, the present chapter is for the most part a description of the improvement of details.

From the experience gained in the construction of the frames of the several steam-driven models, it was decided that the frame for the large aerodrome must consist essentially of two principal parts. First, a rigid backbone was required, extending from the point of attachment of the front wings to the point of attachment of the rear wings; and this backbone, for convenience designated [p165] the “main frame,” must support the second principal part, the “transverse frame,” which formed a cross with the main frame, and at the ends of which the propellers were mounted. While it was necessary that this transverse frame should have considerable rigidity and strength in a vertical direction, yet its main strength and stiffness was required in the horizontal plane for withstanding the thrust of the propellers. It had been possible to construct the frames of the later steam-driven models stiff enough, and at the same time light enough, by the use of properly proportioned steel tubing, but calculation very soon showed that in order to secure sufficient rigidity for the frame of the large aerodrome and at the same time keep the weight within the permissible limit, it would be necessary to depend very largely on guy-wires and to use tubing only for forming the struts against which the guy-wires should act. But this obviously introduced a new series of problems. The extensive system of guy-wires necessary would add materially to the head resistance of the aerodrome, and this might conceivably be so great as to require more propulsive power than would be required for a frame heavier but unincumbered by the head resistance of the wires. It became necessary to consider these problems, but no data were accessible from which the head resistance could be computed with any confidence. The coefficient of resistance for a cylindrical body moving through the air in a direction perpendicular to its length may in general be taken as one-half that of a flat body of the same cross-section; but it was thought very certain that, owing to the fact that tightly stretched wires are in constant vibration when the aerodrome is in the air, the resistance of the wires must be considerably greater than would be calculated from treating them as cylinders having a coefficient of 0.5. Unfortunately, no data on the resistance of vibrating wires were at hand. Before proceeding with the designs for the guying of the frame, therefore, the following brief series of tests was made in November, 1898, on the whirling table, in order to learn approximately the resistance that the proposed system of guy-wires for the large aerodrome would offer:

MEASUREMENTS OF THE RESISTANCE OF GUY-WIRES, USING FRAME ATTACHED TO “BALANCE.”

The last column of these tables is calculated for a coefficient of form equal to 0.5, which has been found to be approximately correct for a rigid cylindrical body.

These tables are not sufficiently extensive to determine accurately the exact resistance that wires of various sizes will offer at given velocities, or to serve as the basis for the deduction of formulæ, and were not made for that purpose. However, from the above data, and the curves plotted in Plate 44, it will be seen that some unexpected results were obtained.

These results are fairly well summarized in the following general statements: First, that the coefficient of resistance increases to some degree as the size of the wire is decreased; second, that in the case of wires of the size which it was expected to use, and at approximately the soaring speed of the aerodrome, the resistance is certainly not greater than 75 per cent, and more probably less than 50 per cent of the resistance encountered by a flat surface of the same projected area; third, that the coefficient of resistance did not seem to be increased by the vibration of the wires. On the contrary, it was noted during the experiments that when they reached a speed which just caused them to “sing,” there was a marked diminution in the resistance. This statement is made, however, with some reserve, for it is probable that the singing of the wires was due to vibration in the horizontal plane, and it is not definitely known what the effect would be of vibration in the vertical plane.

To make the very extensive experiments necessary to determine these propositions conclusively would have required much more time than could at this period be spared from the actual constructional work on the aerodrome. Nevertheless, the data did seem to indicate that it was at least not unwise to employ the extensive system of guying which had been planned in order to give the necessary strength to the frame of the large aerodrome. This plan of construction was, therefore, definitely adopted, and as a result of later experience the system of guying was still further extended.

As the transverse frame had to be made comparatively rigid in order to prevent undue binding of the bearings of the transmission and propeller shafts, it was necessary to make it intrinsically stronger and, therefore, heavier in proportion to its size than the main frame. The main frame, although requiring great strength to enable it to withstand the strains, both torsional and direct, which were imposed upon it by the weights which it supported, did not need excessive rigidity, and could, indeed, be distorted an appreciable amount without danger of any serious effect on the action of the wings or rudder; but even a small amount of distortion in the transverse frame might easily cause such friction at the bearings of the shafts as to absorb fifty per cent or more of the engine power.

In the photographs, Plates 45 to 48, which show the actual condition of the frame on January 31 and February 1, 1900, the letters A, B, C, D, E, F, G, H and I designate parts of the main frame, A and H being the rear and front midrods, respectively, to which the wings were to be attached. B and I are curved extensions of the starboard main tube, the port main tube being exactly similar, and C, D, E, F and G are cross-tubes which connect the midrods to the port and starboard tubes. R is the front main tube of the transverse frame, the rear main tube being exactly similar, and both being connected to the main tubes of the main frame where they cross them. The ends of the main tubes of the [p168] transverse frame are joined together by the “bed plates” L, which are of I-beam section, and have mounted on their outer faces the bearings which support the propeller shafts. At V are bevel gears mounted on the propeller shafts, which are driven by co-acting bevel gears, M, mounted on the outer ends of the transmission shafts, O, the latter being at this point firmly supported in bearings mounted on the inner faces of the bed plates and steadied by the intermediate bearings, N. The two transmission shafts are seen to be not in line, the rotary cylinder engine that was then under construction requiring this arrangement. The bed plates, L, are further stiffened by the brace tubes, K, and the transverse frame is braced against the thrust of the propellers by the tubes J. The four tubes, P, unite at their upper ends to form what was designated as the upper “pyramid,” and the wires, S and T, radiate from its apex to the rear and front, respectively, of the main frame. The lower “pyramid,” on the under side of the frame, also has similar wires running fore and aft. The main portions of both frames are further strengthened by their sub-frames, which merge together, and the main tubes of the main frame are individually stiffened in the vertical plane by a minor system of guying. The scales shown in the photographs are calibrated in metres.

It is to be particularly noted that the midrod, which had heretofore formed the backbone of the main frame, was now made to act merely as a means of attaching the wings to the frame, the main strength of the frame being furnished by the two parallel fifty millimetre tubes which extended the entire length of the frame and which, reinforced by the guy-wires, formed a truss not only more rigid transversely, but also many times stronger in its ability to resist torsional strains than could be secured by a single tube of equal weight. In this plan of constructing the main frame, the pyramids constituted a very important element, for with the guy-wires arranged as they were it was impossible for any portion of the frame to experience a stress which was not transmitted in some way to the pyramids. In the frame, as here shown, these pyramids were formed of tubes 15 mm. in diameter, 0.5 mm. thick, stiffened against buckling under the end pressure by means of the cross-braces, which united them near their midpoints. While the sole function of the upper pyramid was to serve in the system of guying the frame, the lower pyramid not only served a similar purpose, but also provided a means for holding the aerodrome to the launching car in the process of launching it, the clutch-hooks gripping around the short horizontal tube at the apex of the pyramid and thus drawing the “bearing points” of the machine firmly against the uprights on the car. In fact, the particular arrangement of these pyramids was largely determined by this necessity for providing means for holding the aerodrome to the launching car, and the form which seemed best suited to the purpose was duplicated on the upper side of the frame.

The “bearing points” were not attached to the frame at the time these photographs were taken, but are seen leaning against the scales in the foreground of Plate 46. Their position on the frame will be more clearly seen in later photographs, where it will be noted that they were made use of in the more elaborate system of guying which was adopted.

While, in general, the frame at this time seemed to be reasonably stiff and strong, yet it was subjected to a very thorough test by supporting it at different points and suspending from it weights to represent the various parts, such as engine, aviator, wings, rudder and so forth, the deflections which were produced by these weights being carefully noted. It was further tested by subjecting it to vibratory strains, such as it would be likely to meet in actual use. After this the whole frame was tested against torsional strains, such as would be caused by the wind twisting one set of wings more than the other. As a result of these tests it was decided that the frame should be strengthened as far as it was possible to do so without greatly increasing the weight, which even now was found to be rapidly increasing beyond what had been calculated as permissible. The main guy-wires were replaced by heavier and stronger ones, and while these were found to add somewhat to the stiffness of the frame, yet something more seemed necessary to insure safety.

The delay in securing the engine, which had been contracted for with a guarantee that it would be delivered in February, 1899, had become so serious and had delayed the completion of the frame to such an extent that the question of building an exact duplicate of the large machine, but of one-quarter its linear dimensions was being carefully considered at this time, and it was decided to make no further changes in the guying of the large frame until after the small one was built. On account of its smaller size changes could be more readily and cheaply made on it, and the advantages of different methods of guying could be just as well studied. Later, when this was completed, it was found that, with the same system of guying that had been used in the larger frame, the model was so very stiff that it did not require any further strengthening, the smaller scale, of course, accounting for the difference. What was thought to be the best system to follow in strengthening the frame of the large machine was, however, first tried on the smaller one, and it was found that for a very slight increase in weight a very great increase in strength could be obtained. This change in the system of guying consisted essentially of building a “trestle” of tubing at a point on the upper side, midway between the pyramid and the rear end of the frame. One of the former sets of guy-wires which passed to the rear of the frame was then replaced by a set which started at the foot of the rear tubes of the upper pyramid, passed over and was fastened to the trestle, and from there passed to the rear end of the frame at the points where the longer guy-wires from the pyramid had formerly been attached. The [p170] guy-wires on the lower side of the frame, at the rear, were correspondingly changed so that the upper and lower systems should be similar, the wires which started from the main tubes at the foot of the pyramid passing to the bearing points, and from there to the rear end of the frame.

In order to keep the main frame of the large aerodrome as short as possible, it had originally been planned to make the distance between the center of pressure of the front wings and the center of pressure of the rear equal to five metres. When these same proportions were followed in the quarter-size model, it was found that it brought the rear wings so close to the propellers that their lifting effect was certain to be interfered with by the blast of air created by the slip of the propellers. It was therefore decided that all things considered it would be best to increase this distance between the wings, even though this involved an increase in weight, partly on account of the increased amount of tubing, and still more on account of the guy-wires which it would be necessary to add in order to make up for the weakness due to increased length. The large aerodrome frame was accordingly lengthened 2.5 feet (76.2 cm.), and the guy-wire system was changed to that clearly shown by the photographs of July 10, 1902, Plates 49, 50 and 51, the black cross-lines on the background being 50 centimetres apart. From an inspection of these photographs it will be seen that two sets of guy-wires were carried from the upper and lower pyramids, respectively, towards the rear of the frame, the first set being carried to the main tubes at the foot of the “trestle” and the bearing points, and the second set to these same main tubes at the second cross-tube. The sets of wires which started from the feet of the pyramids were carried over the “trestle” on the upper side and the bearing points on the lower side, and both joined to the main tubes at the rear cross-tube. Additional cross-guy-wires for stiffening the frame sideways were added in each of the squares formed by the junction of the cross-tubes with the main tubes. A secondary system of truss guy-wires running over short guy-posts attached to the tubes of the main frame also contributed to the strength and rigidity of the whole.

Although the pyramids had shown no signs of weakness, nevertheless, because of increased strains due to the lengthening of the main frame, it was thought advisable to make them stronger. Instead of the 15-mm. tubing, which had formerly been used, 25-mm. tubing of the same thickness was therefore substituted, and additional cross-braces were added, as will be seen from the photographs, and from the scale drawings in Plates 52, 53 and 54, which show the aerodrome as it was when completed. The numerals attached to these drawings refer to the detail drawings shown in later plates.

In order to secure the proper adjustment of the guy-wires, not only of the frame but of many other parts, notably the wings, propellers and rudder, it was necessary to use a large number of turn-buckles. As almost every wire [p171] required at least one, and in some cases two turn-buckles, the weight represented by this single item rapidly became so formidable as to require serious attention. In the construction of the models, it had been necessary to employ some special turn-buckles in connecting the guy-wires of the wings to their guy-posts in order to secure the minute adjustment of the wires necessary to prevent the wings from being warped and distorted by unequal and improper adjustment. These turn-buckles had been made in the Institution shops, as the very lightest ones which could be secured in the market were from ten to twenty times as heavy as it was necessary for them to be to provide ample strength. In the construction of the large aerodrome, however, the large number required, and the desire to complete the machine at the earliest moment, made it advisable to procure the turn-buckles, if possible, from outside sources, and a very careful search was accordingly made among the various dealers. After much delay some bronze turn-buckles were secured which were very much stronger for their weight than any others on the market, but upon testing them it was found that while they weighed 45 grammes, their average breaking strength was only 593 pounds. Previous experience had shown that turn-buckles which would not break under a less load than 750 pounds could certainly be made to weigh not more than 18 grammes. As even at this time it was realized that at least 100 turn-buckles would be necessary for the entire machine, the excess weight which the heavy turn-buckles would add was felt to be absolutely prohibitory, and the construction of steel turn-buckles was immediately begun in the Institution shops. These turn-buckles were at first made in several sizes, and while some few were at first made “double ended,” most of them were threaded at only one end, the other end being provided with a swivel-hook, or eye. They were at first made of mild steel, the swivel-hooks, in fact, being made of wire nails in order to utilize the head of the nail as a shoulder without the expense of machining rod steel of a size large enough to form the shoulder. It was found, however, that the weak point of this type of turn-buckle was the swivel end, and most of those which were then on hand were made double ended by removing the hook, tapping a left-hand thread into this end of the shank, and fitting a threaded eye-socket in it. The guy-wires themselves were attached to the eyes of the turn-buckles and to the fittings on the frame by twisting loops at the ends of the wires, and although the very greatest difference in the strength of a completed guy-wire may result from the way in which the loops are twisted, yet, after much training, the workmen were taught to twist these very uniformly, following the plan which can be best understood by an inspection of the drawings in Plate 55 which show the loops more clearly than they can be described. After the loops had been properly twisted, soft solder was run all through the twist in order to unite firmly the twists of the wire. Although special grades of wire were found which showed very high tensile strength when the wire was [p172] tested without having loops formed in its ends, yet it appeared that the twisting of these high-grade wires so seriously affected them that in the case of guy-wires with loops at the ends, better final results could be obtained by using softer grades of steel. The wire which was actually found best, after much experiment, was a good grade of Bessemer steel of a medium hardness, which had been “coppered” to prevent rusting. However, even with the softer grades of steel wire, it was found that there were sometimes hard spots in the wire which revealed themselves only upon test, and that when a hard spot occurred in the twisted portion where the loop was formed, the final strength of the completed guy-wire was sometimes only twenty-five per cent of what it should be. The precaution was then taken to subject each of the completed guys to a test strain at least twenty-five per cent greater than it was calculated the wire would have to stand in actual use, so that no accident from defective wires would be likely to occur.

Later on, however, much trouble was caused by the loops in the ends of some of the guy-wires slipping, owing to the giving way of the solder which had been run through the joint, the amount of slipping, while small, being sufficient to alter completely the relative stresses on the various wires, thus causing distortion of the framework itself. In order to avoid this difficulty a new method was devised of attaching the guy-wires to the turn-buckles and to the fittings by which they were carried to the frame. This method consisted in threading the ends of the guy-wires so that they could be inserted directly in the threaded ends of the turn-buckles. The wires when connected in this way to the turn-buckles showed absolutely no slip, and the entire system gained greatly in strength thereby. The only disadvantage which was found in this new method of attaching the guy-wires to their fittings, was that if the wire was bent very close to the fitting, it would break in the screw thread very easily. But since most of the guy-wires when once attached to the machine are always tight, and in fact, under more or less strain, there was in most cases no likelihood of the wires being endangered by being bent close to the fittings. Since the screw threads, which it was necessary to adopt in this new plan of connecting the guy-wires, had to be very much finer than the threads which had been used in the turn-buckles previously constructed, it was necessary to make new turn-buckles, the others being too thin to permit of their being bored out, bushed and re-threaded. The new turn-buckles were made of a much higher grade of steel, and probably represent very nearly the maximum of strength for the minimum of weight possible without the use of some of the very much higher-grade steels which have recently come on the market, but which are exceedingly expensive to work. By means of this improved plan of attaching the wires,42 it [p173] was found possible to gain practically fifty per cent in the strength of the entire system of guy-wires used on the frame.

Many small changes were from time to time made in the various small fittings by which the guy-wires were attached to the frame, nearly all of these fittings having been originally made of a very mild grade of steel owing to the fact that it was so very much easier to work. At the time these fittings were made it was constantly expected that a trial of the aerodrome would be possible very soon, and it seemed necessary to expedite the work as much as possible and avoid the delay involved in using grades of steel that would have been materially harder to work. As is always the case in work of this kind, retrospect shows many instances where what was supposed to be a short cut to results actually proved to be the longest path, but the work as a whole was remarkably free from imperfect parts which necessitated reconstruction.

In the construction of the frames of the models it had been customary to fit the tubing accurately at the joints and to join it permanently together by brazing, as this was not only the lightest form of joint that could be made, but also the most expeditious method consistent with securing a strength of the joint comparable with that of the tubing itself. The construction of the frame by this method of brazing the joints together permanently, offered, however, several serious drawbacks: among them, that when a tube got injured it was a considerable task to replace it, while the brazing of the new tube in place required extreme care to prevent the frame from being warped when completed, as the tube became longer while very hot and contracted after the joint had set. Furthermore, the great heat required destroyed to a considerable degree the desirable qualities due to the tube being “cold drawn,” a reduction of strength of something like 25 per cent being almost inevitable, even when the brazing was most carefully done. It was, therefore, decided that in the construction of the large machine all of the main joints should be made by a system of “thimbles,” and it was planned at first to make these thimbles by brazing short pieces of steel tubing into the proper shapes and angles so that they would accurately fit the tubes which were to be joined. The construction of the thimbles in this manner, however, seemed to involve an excessive amount of work; and, as it was found that very thin castings of aluminum-bronze could be obtained, which would show a tensile strength very nearly as great as steel, it was decided to make up patterns for the thimbles and cast them of aluminum-bronze.

The aluminum-bronze castings were obtained and properly machined to fit the tubes, but when it was attempted to “tin” the interior walls of the thimbles it was found that the solder could not be made to stick to the bronze. As a considerable amount of work had been expended on the machine work of these thimbles much time and effort was spent in attempting to devise “fluxes” [p174] and solders which could be made to work with the aluminum-bronze, but the final result was that the aluminum-bronze thimbles had to be abandoned. They were replaced by similar castings of gun-metal of a slightly heavier section, which at the time were thought to be very suitable for the purpose.

But, in finally assembling the frame after the changes described above had been made, steel thimbles, built up of short pieces of tubing, as had originally been planned, were substituted for the gun-metal thimbles. This change was made not only because of the great increase in strength, but more particularly because many of the gun-metal fittings had been imperfectly constructed, so that it was extremely difficult to align the frame. The steel thimbles, which were made in the Institution shops proved thoroughly satisfactory and gave no trouble of any kind. Many of these thimbles and the method of attaching the guy-wire fittings to them are shown in Plates 56 and 57, as well as in Plate 55.

TRANSVERSE FRAME

It will be recalled from the description of the models Nos. 5 and 6, in Part I, that the position of the line of thrust, with respect to the positions of the center of pressure and center of gravity in the vertical plane was, theoretically, very much better in No. 6 than in No. 5. In designing the large aerodrome, it was desired to reproduce as nearly as possible the relative position of the line of thrust with reference to the center of pressure and center of gravity which existed in No. 6, but for constructional reasons it was found impossible to do so. In fact it appeared that without seriously complicating the construction of the frame it was impossible to raise the line of thrust with respect to the center of gravity materially higher than it was in No. 5. In No. 6 the line of thrust was 12 centimetres above the midrod, this being effected by placing the engines some distance from the boiler, and at the extreme ends of the transverse frame where they were connected directly to the propellers. In the case of the steam engine the weight of the engine proper is a relatively small portion of the entire weight of the power plant, and it is, therefore, possible to put the engine almost anywhere without materially affecting the center of gravity. But where a gas engine is used the engine itself constitutes the greater part of the weight of the power plant, and any raising of the engine, therefore, materially raises the center of gravity of the whole machine. The line of thrust in the large aerodrome was, therefore, practically in the plane of the main frame, and consequently very little higher than the center of gravity.

The use of one engine to drive two propellers mounted at opposite ends of the transverse frame, and in a direction perpendicular to the crank shaft of the engine, necessitates the use of a pair of bevel gears between each of the propeller shafts and the shafts by which the power is conveyed to them from [p175] the engine shaft. Since the efficient transmission of power through bevel gears requires that they be very accurately placed with reference to each other, and maintained very accurately in this position while they are at work, it was necessary to make the transverse frame very rigid, especially at its extreme ends. This was accomplished by the use of what were called “propeller-shaft bed plates.” They are designated by the numeral 27 in Plate 54, and are shown in detail in Plate 58 as of a very deep I-beam section, having very narrow flanges top and bottom, the web of the I-beam furnishing the strength in a vertical direction, while sufficient stiffness laterally was obtained from the flanges, assisted by the brace tubes, which acted as struts between the bed plates and the main tubes of the transverse frame. These struts, while very light, added enormously not only to the lateral stiffness of the propeller bed plates, but furnished for a minimum weight a maximum prevention against twisting of the plates. The propeller-shaft bed plates were originally planned to be made of sheet metal with the flanges brazed to the web. But at the time that they were constructed the pressure of the work was so great in the Institution shops that it was found necessary to have some of the work done outside, and the parties who undertook the construction of these bed plates were unwilling to attempt to braze them up, and accordingly worked them from steel forgings made for the purpose. The expense of this plan of construction proved large and unnecessary, as both previous and later experience proved that it was not only practicable to braze up bed plates more complicated in their design than these, but that equal strength for equal weight could thus be obtained for less than one-quarter the cost of constructing them from solid forgings. Furthermore, where such parts are made from the solid, changes which later tests prove advisable can frequently not be carried out without very serious cost and delay, while with the bed plates formed by brazing less hesitancy is felt in removing parts which are brazed thereto and substituting new parts, or even discarding the bed plates altogether and substituting new ones. Particular emphasis is laid on this point for the reason that much expense and delay would have been avoided had these very expensive propeller-shaft bed plates been discarded as early as 1901 and replaced by others which would have permitted a considerable strengthening of the ball-bearings, which, while strong enough to stand even more power than they were originally designed for, were far too weak to be safe when working under the greatly increased stress due to the very much higher engine power which was later used. Instead of discarding these bed plates then for new ones, they were strengthened by brazing to them crescent-shaped pieces, as shown in the drawings and photographs. This strengthening was made necessary by the larger hole cut in the bed plates for the larger bevel gears. The bed plates for the engine, which are later described, besides other bed plates which were made for other purposes, were all [p176] formed by the use of sheet metal and tubing properly brazed together, and none of them ever gave any trouble.

In the early photographs of the aerodrome frame, especially that of January 31, 1900, Plate 45, it will be noted that the two transmission shafts, which extend from the propeller-shaft bed plates towards the center, are not in line, the port transmission shaft being at the center of the transverse frame, while the starboard shaft is three inches to one side. This arrangement was necessary in order to connect the shafts to the rotary cylinder engine which was being constructed under contract, and which was almost momentarily expected for more than a year after its original promise of delivery on February 28, 1899. Later, when this engine was finally found to be a failure, and the writer constructed the engine in the Institution shops, the starboard transmission shaft was moved over to the center line and the crank shaft of the engine, which was carried through on the center line of the transverse frame, was then connected directly to the inner ends of the transmission shafts.

These shafts, as well as the propeller shafts, were originally constructed of steel tubing 1.5 inches in diameter and 116 of an inch thick, but on account of the increased power of the large engine it was found necessary to increase the thickness of the shafts to 18 of an inch. Difficulty was also found with the tubing of which the shafts were made. This, though not exactly straight when received from the factory, could be pretty accurately straightened in the lathe by exercising proper care, but the moment any real strain was put upon it in the transmission of power, it again went out of shape and caused serious damage to the bearings by whirling, buckling, and so forth. As the skin of the tubing is really the strongest part, owing to the cold-drawing process to which it has been subjected, great care was taken to secure shafts which were sufficiently straight for use without machining, but it was finally found impossible to rely on the unmachined shafts, and all the later shafts for the aerodrome were made by getting tubing a sixty-fourth of an inch thicker than was calculated to be necessary and turning off this extra metal in a lathe.