as simultaneously to connect the condenser with the contiguous chamber and close it to the atmosphere.

The condenser was of the surface form, and the exhaust steam was sent through it, both for the condensing and non-condensing tests, the liquefied steam being weighed on platform scales. The duration of the tests was from 30 to 60 minutes; but weighings were made every five minutes, to prove that the rate of consumption was uniform.

The change froin non-condensing to condensing was made in less than 30 seconds, and the rate of steam consumption changed so quickly, that, after the first 5 minute interval, that rate was uniform. The following is a sample of the water record:

cut-off; 270 revolutions; boiler pressure, 90 lbs. Load: Westinghouse dynamo, 30 ampères and shop shafting.

It should be understood that the 5 minute weighings are only intended as checks to assure us of the practically uniform condition of the whole system. The water rán continuously into one weighing tank until about 400 lbs. were accumulated, when the stream was switched into another weighing barrel, thus allowing the first barrel to be weighed at leisure. Consequently discrepancies in the 5 minute readings are reduced in the final result to a fraction of the amount recorded.

By careful attention, however, it is quite possible to make continuous 5 minute weighings which are very accurate, even taken individually.

For example, the following are three sets of weighings under different conditions.

Under the method described above there is for each non-condensing test a companion condensing test at so nearly the same useful load that the cut-off is always a little shorter for the condensing test than for the non-condensing test; and it so appears upon the plate, which gives the curves of consumption per indicated horse power for 5% clearance.

Several theoretical curves for zero clearance are also given. These are calculated for a Mariotte expansion curve, assuming no cushion and release at 100% of stroke.

It may be observed that the condensing tests invariably show more economy, both for automatic cut-off action and throttling. In the throttling tests a thermometer was placed in the stationary part of the steam chest and another in the steam pipe, so that the superheating due to wire-drawing was determined as indicated on the plate. The steam contained about 3% of water at its entrance to the steam chest, due to the radiation from the steam pipe, as a thermometer at the boiler proved the steam to be about 6° superheated at that point.

The loss due to throttling is clearly shown to be a considerable amount, as shown quite conclusively to my mind by tests made several years ago by the late J. C. Hoadley; but as there appears to be considerable faith expressed at the present time in the ability of throttling regulation to compete with cut-off regulation, I have made these few throttling tests to reinforce Mr. Hoadley's data. The latter are plotted upon the plate, but are not connected by a *I am indebted to the courtesy of Mr. J. C. Woodbury for the results of Mr. Hoadley's tests.

curve. The plate also exhibits the results of Prof. Peabody's tests on the Porter-Allen and Corliss engines at the Massachusetts Institute of Technology. There is also shown the consumption curve for 90 lbs. pressure given in Mr. Emery's paper. The condition of the Buckeye engine was as follows:

1. With the fly-wheel clamped, and 90 lbs. of steam acting on the crank end of the piston, and the forward cylinder head removed, no leak past the piston rings could be detected; but there was a slight leak through the forward inlet port.

2. With the fly-wheel clamped, and either inlet-valve open so that steam at 90 lbs. pressure could run into one end of the cylinder and blow out through the indicator hoie at that end-when the cut-off eccentric was pulled around by hand so as to shut off the entrance of steam to the cylinder, the blow at the indicator cock entirely ceased.

3. With the steam chest cover removed, and the valve at its middle position, 90 lbs. of steam through the open throttle valve, showed itself nowhere except at the lower corner of the forward end of the exhaust or main valve. At this point there was a slight blow of steam. An examination of the valve surfaces found them smooth and polished except at the leaky corner, where there were a few scratches.

4. Under the conditions last described, but with the steam chest cover on, the accumulation of steam in the condenser for one hour was determined to be 35 lbs. I believe that this leakage is greater than what occurs in the actual running of the engine, on two accounts:

1st. As the engine cooled off during the hour of time that the leakage was accumulating in the condenser, the rate of leakage increased, because of the increased percentage of water in the steam chest.

. 2d. When the valve is in rapid motion, the leakage through any small opening is either much reduced or entirely annulled.

On the whole, therefore, I regard the condition of the Buckeye engine as practically perfect regarding leakage, and the excellent. economy shown by the non-condensing 5% clearance curve of consumption on the plate certainly confirms this view. Prof. Peabody assures me that the valves of the Porter-Allen and Corliss engines, tested by him, were also tight, as were Mr. Hoadley's engines.*

* The latter had 10% clearance, but compressed to from half to full boiler pressure.