574 
MESSRS. T. E. THORPE AND J. W. RODGER ON THE RELATIONS 
(j) C increases as the molecular weight increases ; 
(2) C for an iso-compound is less than for the corresponding’ normal compound ; 
(3) C for an allyl compound is less than for the corresponding’ normal propyl 
compound but greater than for the isopropyl compound. 
Exceptions to (1) occur in the case of lowest members of the series of fatty acids, 
aromatic hydrocarbons, and dibromides. Exceptions to (2) and (3) occur in the case 
of the alcohols. 
As to the magnitude of C in passing from one series to another nothing very 
definite can be said. The corresponding iodides, bromides, and chlorides give values 
which are in the order of their molecular weights; the acids and alcohols, however, 
although possessing smaller theoretical molecular weights than the corresponding, 
iodides, give larger values of C, due, doubtless, to the influence of molecular 
complexity. 
As regards the other constants of the formula, n and h are terms connected with 
the temperature variation of rj. It is evideiit, however, from the appended table that 
the magnitudes of these terms are Jiot simply related to the chemical nature of the 
substances. Pentane and isopentane, for example, give different values for these 
constants, although the two experimental curves are almost superposable. 
This is doubtless due partly to the facts — 
(1) That Slotte’s formula does not express the true law of the temperature change 
of the viscosity, and 
(2) That only three observations are used in deducing the formula; but mainly to 
the circumstance 
(3) That the values of n and h are interdependent, so that different pairs of values 
of n and h may be found which give, with the same value of C, practically the same 
viscosity curve, and, from the mode of deducing these constants, the individual values 
of n and h are often affected by influences which fall within the limits of experimental 
error. 
(1) and (2). For short straight curves the formula gives nu’mbers which closely 
agree witli the observed values, the differences exhibiting no regularity. In the case 
of isopropyl chloride, for instance, the calculated values agree with those of observa¬ 
tion to the fourth significant figure—that is, on an average, to 1 part in 2000 ; or 
with a degree of accuracy which is certainl}^ as high as we may suppose the observa¬ 
tions themselves to possess. As soon, however, as the length of the curve increases, and 
the slope begins to vary considerably as the temperature rises—that is, as soon as the 
observed curve commences to deviate to a marked extent from the linear type — the 
formula begins to break down, and the differences vart^ in a regular way, and indicate 
that at low temperatures the calcidated curve is to the right and at high temperatures 
to the left of the observed curve. Attempts were made to obtain a better agreement 
in cases sucli as tins. Professor Henrici, to whom we are indebted for much 
assistance in the mathematical treatment of our results, spent some considerable time 
