FLUIDITY AND THE CHEMICAL COMPOSITION 107
of Rellstab and also of Gucrout, and not in establishing quantita-
tive relationships.
Struck by the fact that metameric substances sometimes have
such widely different viscosities, e.g., isobutyl alcohol 0.03906
,and ethyl ether 0.002345 at 20°, Bruhl (1880) noted that the one
with the higher viscosity generally had the higher boiling-point
and index of refraction. But to this observation Gartenrneister
(1890), testing a large number of substances at 20° or over a
range of temperatures, found numerous exceptions.
It was at this point that Thorpe and Rodger (1894) decided
to make an intensive study of the whole subject of the relation
between the viscosity of liquids and their chemical nature. Their
first care was to work out a method which would give them a far
greater precision of measurement than had been obtained by
many of their predecessors. They then carefully purified
some 87 substances and measured their viscosities from 0°C to
the boiling-point of each substance. An exhaustive search was
then made for a basis of comparison which would bring out the
quantitative connection between the viscosity and the chemical
nature of the liquids. In this search they compared the viscosity
coefficients (17), the "molecular viscosities" [^(Af/p)^], the
"molecular viscosity work" (rjM/p), and in order to make the
comparison under comparable conditions they made the com-
parisons at the boiling temperatures, at "corresponding"
temperatures, at temperatures where the slopes of the viscosity-
temperature curves are equal, and at slopes varying under speci-
fied conditions. They furthermore compared the constants in
the empirical equations which they found to best reproduce the
observed viscosities as a function of the temperature, and they
also compared the temperatures corresponding to a given slope
in the viscosity-temperature curves. Their choice of tempera-
tures of equal slope as a basis of comparison deserves a word in
explanation. They found that on comparing the viscosity
curves of substances which gave the best physico-chemical rela-
tionships at the boiling-point that the general shape of these
curves was the same, or in other words the slopes of the substances
at their boiling-points were practically identical. On the other
hand, alcohols and other substances, which gave little evidence
of physico-chemical relationships, had invariably a different