402 
from a quantitative solution without wash- 
ing it, and, while they must advise the 
elimination of tabulated statements of the 
failures encountered on the road to suc- 
cess, and must, in general, curb a rather 
natural desire to send a four-page reprint 
containing two-pages worth of work to 
one’s friends, they will welcome an in- 
creased exactness of statement of the con- 
trolling conditions of analyses, especially 
when it appears that these factors have 
been intelligently studied. 
It is too obvious to really need mention, 
that not all of the published work is faulty, 
and it is notably true that some of the 
more recent work is of exceptionally high 
character and sets an excellent standard 
toward which analysts in general should 
strive. Reference will be made to only 
two well-known publications—those of 
Noyes, Bray and their associates on the 
development of a scheme of qualitative 
analysis on a scientific and exact basis, and 
the work of Allen, Johnston and Adams on 
the determination of sulphur as barium 
sulphate. In both these there has’ been a 
systematic study of sources of error and, 
where apparently unavoidable errors ap- 
pear, the possibility of correction for them 
has been carefully investigated as well as 
the determination of the special conditions 
under which these errors could be reduced 
to a minimum. Dr. Bray has stated to me 
(and this view is confirmed by Dr. Bétt- 
ger) that possibly the greatest service of 
physical chemistry is the impetus which 
the precise methods which have been 
worked out have given to the more critical 
study of the sources of error and to their 
prevention or correction. This, in their 
opinion, is a greater service than any other 
single attainment in itself. 
It may be admitted that few employed 
as analysts can hope to obtain so complete 
an insight into other fields as is possessed 
SCIENCE 
[N.S. Vou. KXXXV. No. 898 
by Noyes, Bray, Bottger or Johnston, 
and their immediate coworkers. But it is 
not unreasonable to point out that a much 
less thorough knowledge would be of 
enormous assistance and would lead to the 
prevention of another fault in the point of 
view of most analysts; namely, the over- 
looking of the element of compensation of 
errors in their work. Few things would 
do more to bring up the character of 
analytical work than a better appreciation 
of the fallacies involved in ‘‘check an- 
alyses’’ or the false sense of confidence in 
an approximate summation to one hundred 
per cent. Just here, I believe, our teach- 
ers are often at fault. Every student tends 
to glorify his ‘‘check analyses,’’ too often 
even to the destruction of his professional 
integrity. Even when there is no ques- 
tion of honesty involved, there is a blind 
faith in their infallibility, and this is re- 
flected in much of the published work. 
The idea that two analyses carried out 
with practically identical weights of sample 
and equal quantities of reagents may 
““aoree,’’ but be perfectly worthless be- 
cause of inherent errors, never really 
seems to penetrate the mind of many men. 
The fact that so simple a thing as varia- 
tion in the quantities taken for analysis is 
a wise precaution in testing a process is 
also unperceived. I do not mean to say 
that most, or possibly all, teachers do not 
point out these matters, but I do mean to 
say that many students never get a clear 
perception of them, and still less of the idea 
that dependence upon compensation of 
errors in an analytical procedure is quite 
insecure unless the factors governing the 
occurrence of these errors are approxi- 
mately known. I feel sure that these de- 
fects in our teaching would be lessened if 
we were, in general, to pay less attention 
to teaching a variety of methods, as such, 
and more to the careful investigation of a 
