3ARUS.] VISCOSITY OF GASES. 279 
in the results from Table 88 are nearly constant in value, showing 
clearly that the value of 770, inserted into these computations, was too 
low. The nearly constant errors occurring- here do not, therefore, in 
any degree invalidate the law ?/ // =// (l + ad")i, but furnish the best of 
evidence to corroborate it. Indeed, the plan here adopted of calculating 
the constant quantity/^ ) is singularly well adapted to exhibit the 
7/o /o 
true character of the experiments made. 
On the other hand, the right-hand division of Table 91 would show 
that the errors 
x-t 
Vo Vl+ad 1+yd" 
where 29<^xl0 5 <37 is to be introduced, are of a more serious kind 
than before. I have, therefore, contented myself in Table 91 with giving 
the individual values of y. 
Effect of imperfect gaseity. — The fact which most seriously antagonizes 
the relation 
is the occurrence in the case of air and for temperatures below 200°, of 
a mean exponent of (l-\-ad") n 9 which is larger than J. O. E. Meyer's 
data for n lie between % andf ; Puluj's between 0.56 and 0.72; Warburg 
finds w=0.77, v. Obermayer n=0.7Q. Holman's elegant method leads 
to data which lie decidedly above 0.75, and which are of such a kind 
as to induce him to discard the exponential form of function altogether. 
Fully cognizant of the purposes Mr. Holman had in view in represent- 
ing his results by a series of ascending powers of 0", I nevertheless 
think that at the present stage of research a conservative policy is 
wiser. Results lying within an interval of only 200° above 0° O. can 
not be depended upon as furnishing a definite or final critique. More- 
over, a formula which fails when tested by exterpolation has no further 
interest than the special one to which the author has applied it. The 
expansion of- 7 ?- in terms of a limited number of powers of Q" alter- 
nately opposite in sign will lose all significance when applied beyond 
the interval of observation, and if the interval be small the use of such 
a formula scarcely justifies the labor involved in the computation. I 
have, therefore, preferred to retain the older formula 
and have endeavored to ascertain whether reasons might not be found in 
virtue of which n would tend to increase in proportion as low tempera- 
tures (0° C.) are approached. My experiments refer principally to 
temperatures above 400°, temperatures at which the conditions of perfect 
gaseity may be more justifiably accepted a priori and within which the 
true law of gases may be more clearly discerned. Dissociation is of 
(933) 
