In reducing these observations, the following corrections were involved : a correction of O002 gramme due to imperfect exhaustion of the A flask, while being weighed as empty ; the correction due to the neck of the flask between the mark and the stopcock not being at temperature T amounted to '000 14 gramme and was practically negligible; the volume of the flask, which only required to be corrected for temperature, was 314-398 c.c.
Experiments 1, 2 were made with liquid oxygen taken to be. at temperature 90 0> 5 absolute. The experiments 3, 4, 5, 6 were made in one and the same sample of liquid air, with rising temperature. For these temperatures, obviously only a few degrees below the boiling- point of oxygen, the ordinary gaseous laws may be held to apply, in order to determine their values. Thus we may employ the formula
90-5
T = v
256-833'
where 256-833 is the mean of the volumes of Nos. 1, 2, to get the temperature of the last four experiments. The values thence obtained are entered in column T.
The first two experiments made with liquid oxygen give a ratio of the nitrogen densities from my own values of 3 -088, the absolute temperature ratio being 3 '01 7 ; my values for the ratio of the oxygen densities for the same range of temperature being 3'091 as previously deduced. We may safely assume that if the density of nitrogen, jvere observed at its boiling-point it would deviate asnmch from the" ordinary gaseous laws as oxygen. Further, the specific volume of nitrogen at its boiling-point of 78 absolute, would from the above formula be 221-3 as compared with 226'2, the similar value found for oxygen.
The general inference to be drawn from these preliminary experi- ments is that reliable vapour densities may be determined at very low temperatures. There seems to be no reason why the vapour density of hydrogen at its boiling-point should not be accurately ascertained ; only, as in this case the internal pressure in the weighing flask would amount to nearly 15 atmospheres, it would be advisable to construct the flask of some metal or alloy. A flask of the size used in the oxygen experiments filled with the vapour of hydrogen at its boiling-point would be equivalent in weight to between 4 and 5 litres of hydrogen at the ordinary temperature and pressure, and such an amount of material ought to give density results at the boiling-point of hydrogen of considerable exactness, notwithstanding the great manipulative difficulties that would necessarily be involved in the execution of such a determination at 21 absolute.
