the reduction of the friction of the principal bearings, as the load was increased, by an increase in the rate of oil supply caused by the more vigorous agitation or knocking about of the crank piu and main shaft in their bearings. I held that such agitation pumped the oil out of oil cups by creating a more active suction. Since the meeting I have made some special experiments to test this hypothesis, and I find that reduced friction does not result from an increased supply of oil where the feed is restricted to such an extent as would make my hypothesis tenable. But I find that vigorous agitation or jerking of a bearing causes a considerable reduction of friction without any change in the quantity of lubricant present upon the bearing surfaces.

I therefore adopt this latter hypothesis, and think it quite probable that it can explain the reduction of friction, with increase of load, since we see by the results in Table E that, even allowing Morin's laws to apply, only a small reduction in the coefficient of friction is required to change the relative order of the values for friction for the three horse-powers, so as to make the greater power give the least friction, allowance being made for the accidental variation of friction being in opposite direction in consecutive

tests.

Mr. Geo. Schuhmann.-I would like to ask Prof. Thurston how the friction of the valve and of the eccentric was determined separately. It appears to me that disconnecting the valve from the eccentric leaves the latter without its load, and consequently there must be less friction of the eccentric strap than when the load is on. To simply run the valve under boiler pressure, with no expansion taking place in the cylinder after the valve has cut off steam (I presume the piston was blocked while engine was run with belt), certainly does not give the exact amount of power required to drive an unbalanced valve when the engine is working under ordinary conditions. I still believe that the only exact way to determine the power required to drive a valve is the plan suggested by me during the discussion of Mr. Giddings' paper on "Valve Dynamometer," Vol. VII., page 642.

Prof. Denton.-I would like to add another point. The engine tested by Prof. Thurston is a link engine; those indicator cards exhibited in the paper are taken at very short cut-off. My experiments with the apparatus described in my paper on the friction of piston packing rings have brought out the fact that even when a considerable tension is upon the piston ring to force it out

against the bore of the cylinder, when full steam is let in on the inside of the cylinder, the friction goes to zero instantly. It is due to the water that is in that steam when it is freshly let in. While the steam is being admitted, the water being in large quantities, is superior to any oil. Furthermore, when the engine is run at full stroke, and the speed of the piston is increased over a range of from 35 to 150 revolutions a minute, we do not have to change the feed of oil at all for 150 revolutions from what we find sufficient at 36 revolutions, showing that the large quantity of steam filling the cylinder at three-fourths cut-off affords a lubrication additional to the oil.

Mr. W. F. Mattes.-I do not wish to challenge Prof. Denton's oil-pump theory, in reference to this apparent paradox, beyond pointing out that its application must be confined to bearings working under reasonable pressures. In practice, as loads are increased, we frequently reach a point where the journals begin to heat. It is very probable, however, that in many such cases the increased rate of lubrication may still prevail, while the effect is neutralized by distortion of the bearing surfaces. I have been convinced, by my own observation, that actual distortion frequently occurs where it has not been suspected, and that the bearing surfaces, upon which our calculations have been based, are thereby sensibly reduced.

Mr. W. E. Crane.-As to Prof. Denton saying that water in steam was a lubricant, I knew of an engine thirty inches diameter of cylinder and five foot stroke, which was supplied with steam through a long pipe ten inches in diameter. This pipe was left uncovered for some three or four weeks. During that time it was impossible to keep the cylinder oiled so that the valves would work smoothly. After the pipe was covered it ran without any trouble, with a quart of oil per day, until it became necessary to change the pipe, and this pipe was also uncovered for some little length of time, and we had the same trouble with keeping the cylinder lubricated. After the pipe was covered there was no further trouble. Of course, when the pipe was uncovered there was greater condensation and a greater amount of water with the steam. In case where a cylinder will work with less oil when the steam follows long than where it is cut off short, if it were not for cylinder re-evaporation there would not be that difference.

It has also been remarked in the discussion on piston packing rings that the steam surfaces would work without oil. It is true

that engines have been run for years without oil in the cylinder Perhaps a homely illustration will show what oil will do on steam surfaces.

Our old locomotives were built with a slide valve for a throttle, and there was also a cup put in top of dome, with a pipe from the same leading into the steam pipe, and it was the fireman's duty before building a fire to put some oil in this cup which went down on the throttle.

A certain engineer had an engine assigned him, but not knowing whether the oil would go into the steam pipe or boiler, would not allow the fireman to put in any oil through fear of causing the boiler to foam. The fireman having all the switching to do was anxious that the throttle should work easier, so one morning he poured a small amount of oil into the prohibited cup. The ease with which the throttle worked satisfied him where the oil went, but he did not say anything. It was also his duty to make up the train, and the engineer had nothing to do till it was time to

start.

The conductor gave the signal and the engineer got his foot braced in his old way, took hold of the throttle with both hands; the resistance due to friction having been removed, the throttle lever straightened out and he went over into the tender.

Prof. Thurston.-Two interesting points have come up in the course of this discussion, but I think I can say all that seems to be necessary in a very few words. In regard to the device presented by Prof. Denton for measuring packing-ring friction, which is the first point under discussion, I have been very much interested in it. It strikes me as a very ingenious thing, and likely to be very useful. I should think the results would probably give us some facts and some figures which would be of great value. It is one of the prettiest things I have seen in a long time, and I think the results reported to-night are exceedingly interesting.

In regard to the pumping action to which reference was made during this discussion, and at an earlier one, I should think very likely, under the stated conditions, that it would take place and affect the action of the engine. I should think where the friction is found to be less with heavy loads than light that that action may be at the bottom of the "paradoxical" behavior, but it rarely occurs in the cases that I have investigated that the friction had fallen with increase of load. If we could detect any difference it was a little greater. I have usually attributed the observable dif

ference to differences in the general lubrication of the machine. My own experience has been similar to that of others taking part in the discussion-that it is next to impossible to lubricate an engine in such manner that the lubrication shall be equally effective for any ten seconds successively; and I have supposed that the irregularities cropping out in this whole series of experiments have been largely due to difference in lubrication at various times. If you read the report of the experiments of Beauchamp Tower I think you will see that the rate of speed of an engine, under various conditions, does considerably affect the efficiency of lubrication. Where a bearing is loose I have no doubt there is that pumping action which is described, and I have no doubt that at times it is very effective. I would not at all dispute the hypothesis. I was asked how we distinguished between the friction of the valves and the friction of the eccentrics and strap. That is very easily done by driving the engine, as we are accustomed to do, by means of extraneous power through a transmitting dynamometer, observing the amount of friction when the valve and its connections were complete; then disconnecting the valve from the stem and running all the rest. Then by throwing off the stem we get the friction of the strap. By throwing that off we get approximately the friction of the engine dismantled of its valve gear. It was suggested that no law was discovered, so far as could be observed in this case. I am not at all sure that it can be. I have never supposed that a law, in a scientific sense, could be found. What we were trying to find was, not the law exactly, but the facts; and we have actually solved the problem which presented itself originally. The question being whether the assumption made by those who have written on the theory of the engine that the added friction varied directly as the load, was right; that question has been very well settled. The irregularity of the results found in these cases, and probably in all cases of investiga. tion, is not surprising, and was fully anticipated. It is not of the slightest consequence, in view of the nature of the work, We have got the line of our stroke of lightning and we find that it is a line of pretty definite path. The variation is not great and is not of the kind that was anticipated by the old authorities. The final result then showed that the practice of engineers has been correct, and that the theories of the text-books have not been prac tically correct. We were right in assuming the resistance to be about constant at all times.

A remark was made about the distortion of the bearings, and it was said that that would sometimes account for the irregularities of the results-I have no doubt of that. I had charge, on my watch, of the machinery of the old iron-clad "Dictator" for about a year, once. There we had a main journal twenty-one inches in diameter and originally twenty-six inches long. I remember my first remark on looking over the machinery was that that seemed a pretty short journal for the kind of work to be done. We went to sea with that engine, and at the end of the first trip we were compelled to file up our journal and reset our brasses; and we found we had cut down the journal about a quarter of an inch in the run from New York to Fortress Monroe. We lengthened out the journal to thirty-two inches, and made another trial with very similar results. Finally, we put in a spring bearing back of the main crank shaft-a bearing two feet long-giving us a total length of between fifty-five and sixty inches. Still the engine gave us a great deal of trouble. The reason of that, I have not doubt, was that the shaft had more work on it than it could do without springing; and if we had made it half a mile long it would have still had the heaviest load at the point nearest the crank pin. It always did heat, and probably always would, unless we had so flooded that journal with oil that the shaft could never break through the oil. That was an experiment of the greatest value to me, and I have since always been careful to see that my journals have ample diameter as well as ample length, and especially in critical cases to see that the shaft was supported on both sides of the crank, instead of only on one, as in that case. I believe I have met all the points which have been brought up this evening.

I want to add also that the suggested differences in frictional packings, which was brought up in discussion of the paper on piston packing rings, has no bearing at all upon the results attained in these investigations. The same engine always has the same ring and keeps on using that ring, and, of course, differences of rings have nothing to do with the variation in methods of friction.