THE VOYAGE OF II.M.S. CHALLENGER. 
$3 
1 Id .1 mi + 0*04 instead of +0*033, which I now think comes nearer the truth. To 
^ S - 0*007 corresponds Ax — — 0*005 ; hence if I had used 0*033, the constant 
• in x 1 — \ would have become = — 0*042. The correction is insignificant ; butwewill 
h «i»t it and formulate the net result of our inquiry by saying (in reference to the 315 
cases considered) that 
X 1 — x=-0*042±S 
where S is a variable quantity of which the chances are even that it is less or greater 
t: m 0*00, and about 8 against 2 that it is loss than 0*12. It would not be fair to charge 
the whole of cither term against Mr. Buchanan as representing an error in his specific 
_ ivitv determination. My analytical values x must be charged with a probable error 
of (my) ± 1 )() th of their value, i.e., of about ±0*01, and part of the constant term may be 
owing to my having unwittingly changed my unit for x when (a considerable time after 
the Challenger water analyses had been made) I analysed my standard waters for the 
specific gravity research. I fed sure in my mind that this latter error could not amount 
to more than ±0*02 at the outside. Assuming it to be negative, the error in the 
numbers x' would be reduced to a(x') = — 0*02±0*05 as an estimate of its “probable” 
value. Only the variable term affects the oceanographic applications; translated into 
a difference of specilic gravity it amounts to about ±0*07, which is a little less than 
the value which resulted from my hydrometer experiments. 
Tb< oceanographic significance of his specific gravities will be discussed by Mr. 
Buchanan himself. 
1 am indebted to Mr. Thomas Barbour for the valuable assistance he has given me in 
iriying out the specific gravity research, and in the tedious calculations involved in it 
and in the construction of the tables. 
