On the Distribution of Temperature in the Gulf Stream. 33 
This accordance is satisfactory whether viewed in relation to 
the probability of recognising the band in passing rath Sak or 
partly for nantical reasons, and also as giving a nearer approach to 
equilibrium than the winter 
The differences in the temperature of the whole mass of water 
at the same season of different years are often more considerable 
than the difference in distribution. 
The second test of the probable accuracy of determination of 
the several principal maxima and minima was by a comparison of 
the independent determination of the maximum and minimum 
points in the curves of distribution at the same depth, correspond- 
ing to various depths from the surface. It was first established, 
by a general induction, that all the points of maximum and min- 
imum, except the “cold wall” minimum and axis maximum, are 
probably, as a rule, vertically over each other. Next the curve 
was found, by which the recession of the first minimum and 
Maximum from the shore, as the depth increased, could be repre- 
Sented. The differences then, from the mean curve of recession, 
for the first two points, and from the vertical line or average posi- 
tion for the other points, gave the probable discrepancy of deter- 
Mination, It would be out of place here to give all these labored 
details. This discussion gives as might be expected, smaller 
probable errors than the other ; for this takes in accidental errors 
only, and that includes real changes. ‘The mean probable error 
The corresponding results for all the sections are given in Table 
No. 1 and show on the average less than one mile of uncertainty 
for the mean determination of the first or “‘ cold wall” minimum ; 
two miles and a half for the first or axis maximum, and the secon 
minimum between them ; and four miles for the next three points, 
and about eight and a half for the fourth minimum, which was. 
sown on but three of the sections. 3 
‘he Hatteras section presents, as before remarked, the most 
Considerable discrepancies in its results, incident, most probably, 
‘0 the nature of the phenomena themselves in thatregion. = 
sow Senres, Vol. XXI, No, 61.—Jan, 1856. 6 
