PHYSICS: W. P. WHITE 345 
into account the result brings A v up to its value for silica glass, within the 
uncertainty of the various data. 
In order to interpret these results it is necessary to make allowance for the 
fact that according to the Kinetic Theory, as usually modified to fit the be- 
havior of specific heats at low temperatures, A v for these oxides is appreciably 
below 5.96 even at 1300°. Any high-temperature variation must appear as a 
variation from the value calculated from the low- temperature results. Nernst, 2 
using the Nernst-Lindemann formula, has derived a special formula for the 
atomic heat of silica glass based on determinations from —247° to +283°, 
and this formula, slightly modified to agree with the present results at 283°, at 
which temperature Nernst had used an approximated value of Magnus', has 
been used to get the values given in line 5 of table 2. 
These values, for 1100° and 1300°, are below the values for cristobalite by 
more than the probable error, which indicates that unless cristobalite changes 
TABLE 2 
Atomic Heats. Ap 
300° 
500° 
700° 
900° 
1000° 
1100° 
1300° 
4.95 
5.35 
5.58 
5.75 
Quartz 
5.46 
5.66 
5.55 
5.67 
5.72 
5.78 
5.86 
Silica glass, formula, A v 
4.95 
5.35 
5.55 
5.67 
5.71 
5.75 
5.80 
Albite 
5.10 
5.46 
5.71 
5.91 
5.97 
5.09 
5.47 
5.72 
5.86 
5.92 
5.22 
5.58 
5.82 
6.04 
6.14 
6.31 
6.82 
Platinum, A v 
6.45 
6.79 
6.84 
very considerably its rate and probable direction of volume change near 10U0°, 
A v for it is increasing with temperature above the accepted theoretical value. 
A j; for silica glass is already well above the theoretical curve at 900°, and is 
probably still further above it at higher temperatures. Unfortunately, this 
glass could not be continuously heated above 900° without danger of crystalliz- 
ing. Unlike nearly all the other series the original values for cristobalite 
were obviously not on a smooth curve, and were smoothed before being used 
to get the true heats. But none of the changes was above 1 per mille, the 
permitted accidental error. A very different plausible alteration gave iden- 
tical results, and none which seemed reasonable gave a value at 1300° lower 
than 5.83 obtained from the unsmoothed data, and even this value is above 
the computed curve. For this curve the Debye formula, and even the original 
Einstein formula, give above 700° results practically identical with those in 
the table, so the conclusions are not dependent on the particular formula used. 
Since the theory applies to solid bodies only, and silica glass, being amorphous, 
has some of the properties of a liquid, the results for cristobalite are more sig- 
