316 Scientifie Intelligence. 
It is surprising how few substances fulfill the conditions re- 
quired in an accurate atomic weight determination, rarely more 
than one compound of each elementary substance ; and for some 
mmon e 
one. After the process has once been worked out, the actual ana- — 
lytical work is degen nave accomplished and the accidental 
errors involved are comparatively small; and when the constant 
errors have all a eliminated t S detinitonise of the results is 
hb : 
the writer, while his earlier work.on crystalline alloys left an im- 
pression of the possible oo (under saben conditions) 
baleened. a result we should expect if a variation were petit 
dle and as was predicted in the memoirs that have been so often _ 
cited. On the ennorins * the writer has found that when ee i 
it, and until the requisite facts have been cata bliaied: thats is ne 
sufficient ground for a positive opinion. But, although it mus 
be admitted that the atomic theory is the only basis on which : 
_ consistent philosophy of chemistry can at present be built, the 
writer must confess that he i is rather drawn to bent view of nature 
namical ge 2 
He aT ese 
PF oOPeORaRy « 
as ede which regards the atomic theory as only a oe exp 
dient for representing: the facts os chemistry to the — 
2. On the Spectrum of B —The position of Heol yiliom FS 
os the elements still remains : doubtful. Nilson and Petterson 
it as a tr riad with an atomic gan of 13°8 sac a 475 
toed 
