FROM SELECTED DAYS DURING THE SEVEN YEARS 1898 TO 1904. 
331 
potential gradient, is drawn on a more open scale than the curves of fig. 2; the light 
curve representing barometric pressure is drawn on a scale such as to make its 
apparent range nearly equal to that of the other curve. 
Whilst the i esemblance is very striking, as between two different elements, there 
are differences which seem of a fundamental character. During the forenoon the 
curves are nearly in the same phase. The barometer curve lags a little, but very 
little, behind the other. In the afternoon, however, the lag in the barometer curve 
becomes conspicuous, amounting to about two hours at the times of the afternoon 
maximum and minimum. 
This change in apparent lag throughout the day does not seem to have been 
noticed previously, but it persists with wonderful regularity throughout the year. 
Treating the maxima and minima as occurring at the exact hours when the 
algebraically greatest and least hourly values occur, I find that the hour of the 
forenoon minimum is the same for the two elements in every month from January 
to October; the mean lag for the barometric pressure for the 12 months comes to 
A of an hour , but the exceptionally early hour of the potential minimum in October 
is responsible for more than half this difference. 
In the case of the forenoon maximum there is exact agreement in the hour in 
eight months, and the mean lag in barometric pressure for the 12 months is only 
^ hour. 
December is the only month when the afternoon minima accord in time, and the 
mean lag for the barometric pressure is l\ hours. The corresponding quantity for 
the afternoon maxima is 2^ hours. 
If the i elation is a case of cause and effect, the fact that it is the barometer curve 
that lags relative to the other would naturally lead one to regard the potential 
variation rather as the cause than the effect. 
§ 28. Y\ lien we consider the variation from month to month in the amplitudes of 
the diurnal inequalities other differences appear. This will be seen on comparing 
Table XVII. with Tables III. and IV. The first column in Table XVII. gives the 
range, or difference between greatest and least hourly values, in the diurnal 
inequality of barometric pressure; the second column the numerical sum of the 
-M hourly differences from the mean for the day. The corresponding potential 
quantities will be found in the two last columns of Table III. The ranges and the 
sum of the differences show analogous phenomena, but the latter are the better for 
comparison, because they give a smoother annual variation when the number of years 
considered is limited. 
Comparing, then, the second column of Table XVII. and the last column of 
fable III, we see that in B (barometric pressure) the values for the winter months 
November to February are decidedly the least, and the values for April, May, June, 
and September the largest. In P (potential gradient), on the other hand, the largest 
values occui in December, February, and March, and these are much in excess of the 
2 it 2 
