Specific gravity determinations, although the results lack uniformity, show differences thought to be sufficiently pronounced to indicate that in general the density of the metal is materially diminished in the vicinity of the place of rupture of tensile specimens, and that this diminution takes place in the different grades of steel, in bars ruptured under different conditions of temperature, stress, and contraction of area.

The specific gravity of the truncated ogival ends of fractured tensile specimens has been found .2 per cent lighter than samples from the original hot-rolled bars. Strain caused by the application of a stress inferior to the elastic limit appears to take place at once. No difference in the deflection of a transversely loaded shaft was detected at the highest speed experimented with when the stresses changed from tension to compression in one forty-fifth of a second, the shaft being run at atmospheric temperature, and, when under the highest load, with a maximum fiber stress of 50,000 lbs. per square inch. Common experience has shown that the full effect of a load superior to the elastic limit is not immediately felt in the elongation of a ductile metal, which appears to be true also at higher temperatures. It is difficult to make a direct comparison of the rate of flow at different temperatures, as the elastic limit and tensile strength are changing in the meantime. A number of examples point to the conclusion that a more sluggish slower rate of flow may take place at high than at low temperatures. A specimen of steel of .97 per cent carbon was observed for a period of three hours, during which the elongation continued at nearly a uniform rate of speed, stretching about .0008 of its length each five minutes, under a stress of 20,000 lbs. per square inch. The initial and final temperatures of the test were 1,170 and 1,189 degrees respectively. Another example furnished by the same grade of metal, in which the observations were limited to one hour, showed the mean rate of flow to be .0015 per unit of length per five minutes, with a slightly accelerating tendency. The stress was 5,000 lbs. per square inch, and the initial and final temperatures 1,505 and 1,494 degrees respectively.

Heretofore we have been considering the effect of higher temperatures upon the strength of the metal exposed to simple tensile

stresses.

Experiments were made with riveted joints, in steel boiler plates,

at temperatures from 70 to 700 degrees, over which range the results of the tensile tests of the plain bars were corroborated. Riveted joints at 200 degrees showed less strength than when cold; at 250 degrees and higher temperatures the strength exceeded the cold tests, and when overstrained, approaching the limit of rupture, at 400 to 500 degrees there was found when completing the test cold an increase of strength over the duplicate cold test made in the ordinary manner. A single riveted butt joint tested at 500 degrees ruptured with 81,050 lbs. per square inch on the net section of plate, whereas the corre sponding joint tested cold failed in the same manner with 65,000 lbs. per square inch. Another joint at 500 degrees reached 119,980 lbs. per square inch compression on the bearing surface of the rivets. Still another joint, which was strained at 500 degrees, then cooled to 150 degrees and ruptured, sustained 137,110 lbs. per square inch compression on the bearing surface of the rivets.

As none of these joints failed by direct crushing of the metal at the bearing surfaces, without defining the crushing limits under these conditions of test, we are enabled to say the metal is capable of sus taining very high compressive stresses in this zone of temperature.

Rivets which sheared cold at 40,000 to 41,000 lbs. per square inch at 300 degrees sheared at 46,000 lbs. per square inch; and at 600 degrees, the highest temperature at which joints were ruptured failing in this manner, the shearing strength was 42,130 lbs. per square inch.

The internal strains in some oil-tempered and annealed steel cylinders have been investigated. The cylinders were numbered 7, 8, and 21.

The salient features of the investigation were that cylinder No. 7, which was oil-tempered, annealed, and then retempered, was found to have the entire surface metal, exterior bore, and ends in an initial state of compression, and the interior of the mass in a state of tension. The maximum stresses found were 47,161 lbs. per square inch compression, and 938 lbs. per square inch tension. The latter stress not representing, however, the maximum amount which was in the cylinder when entire, the inaccessibility of the tensile metal at the time preventing that value being ascertained. The strains in cylinders Nos. 8 and 21 were of small magnitude, showing the final process of annealing to have been very efficient. The middle section of cylinder

.

No. 8 was cut into two slices and each re-treated at Watertown Arsenal: the earlier treatment was done at the steel works. Slice No. 1 was heated cherry-red and quenched from the bore with oil at 65 degrees Fahr. Slice No. 2 was heated bright-red and cooled from the bore with water at 80 degrees. The maximum stresses in slice No. 1, computed from the strains which were released when the slice was cut into detached concentric rings, were 18,984 lbs. per square inch tension, and 34,669 lbs. per square inch compression, the compressive strains being next the bore. The maximum stresses in slice No. 2 were 50,814 lbs. per square inch tension, and 59,060 lbs. per square inch compression, a total range of 109,874 lbs. per square inch in the same piece of metal. The compressive stress exceeded the primitive elastic limit of the metal. Even this is not all, for the observed stresses were the means for their respective rings, the rings which were cut apart radially showing additional strains not released until then. The inside ring of slice No. 1 was turned down in the lathe, reducing its exterior diameter in successive stages each, until its thickness had been reduced from .25" to .05" thick; in the meantime the bore continued to expand at each operation until the stress corresponding to the strain released was 50,720 lbs. per square inch, against 34,669 lbs. per square inch displayed by the entire ring. The ring was then turned down to .025" thickness and cut apart radially, whereupon the ends opened .147," thus showing that a difference of intensity of stresses remained in this thin ring. Retrogression of intensity of strains occurs very rapidly when departing from the quenched surfaces. Observations were made on the persistence of internal strains under higher temperatures. Detached rings from cylinder No. 7 were cut apart at one side, the ends wedged apart, and the rings then exposed to the annealing effect of different temperatures. The wedge was driven until the elastic limit was exceeded, as shown by the permanent set in the chord measurement across the cut. In this condition the ring was under the maximum internal strains which it was capable of receiving at atmospheric temperature. The additional permanent sets observed on the chord measurement indicated the release of internal strains after exposure to higher temperature. Strains were released at 428 degrees, and increased sets found as higher temperatures were employed. The highest temperature reached was about 1,450 degrees. This temperature did not entirely eliminate

the internal strains, the restoration in chord measurement which followed the removal of the wedge after this temperature showed the metal had an appreciable elastic limit at that temperature, although a low one. This same ring was next heated cherry-red and quenched in oil. Now, subjecting it to the temperature 410 degrees, wedged apart substantially the same amount as before, and the permanent set found was over six times the magnitude of the set after heating to nearly the same temperature in the first instance. This remarkable difference in the persistence of internal strains displayed by the ring before and after the last retempering demands further investigation in order to ascertain the influence of intervening periods of time, of different initial states of hardness, and different methods of tempering and hardening.

Specific gravity determinations with sectors from cylinder No. 7 (the pieces were small, weighing in air about 33 grammes each) showed, after heating cherry-red and quenching in oil, a slight increase in density, and when quenched in water from the same temperature a decided loss in density. Again heating cherry-red, and reversing the pieces in the quenching fluids, the same differences were displayed as before, that is, quenching in oil caused an increase in density, quenching in water a decrease in density. Heating nearly white hot and quenching in oil caused a decrease in density, as the water had done at lower temperature. As similar treatment is found to cause in different specimens both internal strains and changes in density, these two features may be regarded as correlated functions.

From what has just been said we see that internal strains are released by elevation of temperature, and the extent to which they are released depends upon the temperature reached. Earlier remarks stated that the elastic limit diminishes with increase of temperature, therefore we infer that strains in excess of the elastic limit at the annealing temperature are released by that temperature, and complete elimination of internal strains would therefore require a temperature at which there was practically no elastic limit.

Phenomena attending the over-straining and alternate straining of iron and steel are under investigation. It appears that certain steel bars which originally possess an equality of elastic limits under tensile and compressive stresses, when loaded beyond the elastic limit in either direction, lose in the elastic limit in the opposite direction.

The loss has been found very serious, amounting in some cases to almost complete elimination of the elastic limit in the opposite direction to the over-straining load. Thus, a metal having elastic limits under tension and compression each 50,000 lbs. per square inch would have a total range of stress of 100,000 lbs. per square inch before overloading. Exceeding the tensile elastic limit, say, 1,000 lbs. per square inch, and there results a loss in the compression elastic limit, so that the total range of stress within the limits of perfect elasticity is now little if any above 51,000 lbs. per square inch. Bars over-strained in this manner have been annealed at different temperatures, from 1,180 down to 278 degrees.

The equality of elastic limits was measurably restored even by the lowest annealing temperature. Under the higher temperatures the restoration was nearly or quite complete.

MEETING 398.

Combination Voltmeter and Ammeter for Electrical Measurements.

BY MR. ANTHONY C. WHITE.

The 398th meeting of the SoCIETY OF ARTS was held at the Institute on Thursday, January 23d, at 8 P. M., Mr. G. W. Blodgett in the chair.

After the reading of the records of the previous meeting, the chairman introduced Mr. Anthony C. White, of the Bell Telephone Co., who read a paper on a "Combination Voltmeter and Ammeter for Electrical Measurements."

Mr. WHITE said: The phenominal growth of telegraphy, telephony, and electric lighting during the past few years has rendered more and more imperative the demand for accurate and commercial apparatus for measuring the various quantities involved in each of these branches of electrical engineering.