OF THE PARAFFINS AND THEIR MONOHALOGEN DERIVATIVES. 
11 
dozen figures next the vibrating tube are omitted. This was generally done, as 
they were almost always found to be irregular and less distinct than the rest. 
Table I. 
Scale 
reading. 
Half wave¬ 
length. 
Scale 
reading. 
Half wave¬ 
length. 
Scale 
reading. 
Half wave¬ 
length. 
155-5 
486-0 
25-1 
815-9 
25-1 
180-6 
25-1 
511-2 
25-2 
841-3 
25-4 
205-6 
25-0 
536-9 
25-7 i 
866-6 
25-3 
2.31-4 
25-8 
562-2 
25-3 1 
892-2 
25-6 
257-0 
25-6 
587-3 
25-1 
917-6 
25-4 
282-4 
25-4 
612-8 
25-5 
943-0 
25-4 
308-3 
25-9 
638-2 
25-4 
968-3 
25-3 
333-4 
25-1 
663-6 
25-4 
993-4 
25-1 
358-7 
25-3 
688-8 
25-2 
1019-0 
25-6 
384-2 
25-5 
714-0 
25-2 
1044-6 
25-6 
409-8 
25-6 
739-7 
25-7 
1070-2 
25-6 
434-9 
25-1 
765-5 
25-8 
1095-4 
25-2 
460-9 
26-0 
790-8 
25-3 
The mean value for the half wave-length is 25’399, and it will be seen that no 
single measurement differs from this by more than six-tenths of a millimetre. In two 
or three sets where the figures were poor the divergence from the mean reached 
as much as a millimetre, but was never greater. In some of the propyl chloride 
experiments it was not more than a quarter of a millimetre. 
The method of calculation of the mean was to divide the readings of the nodes 
into two equal sections, subtract each reading in the first section from the corre¬ 
sponding one in the second, take the mean of these differences and divide by the 
number of half wave-lengths between the first readings of the two sections. 
§ 3. The Vapour Density Apparatus. 
The ordinary methods for determining vapour densities are not very suitable for an 
investigation such as this. Hofmann’s, Victor Meyer’s and Dumas’ are scarcely 
accurate enough, and the two latter are not applicable without modification to 
pressures other than that of the atmosphere. Regnault’s, though very accurate, 
would take too much time, when so many determinations have to be made. 
As the vapour densities are only required at the temperature of the room, the 
conditions are much simplified, and I have devised a form of apparatus, using 
Hofmann’s principle, which gives results concordant to per cent, without any 
great expenditure of time. Doubtless with greater precautions for securing uniformity 
of temperature higher accuracy might be obtained, but an error of one part in a 
thousand is well within the experimental errors of the rest of the work. 
c 2 
