134 THE JOURNAL OF GEOLOGY. 
is the amount of rise or fall in a fixed interval, and corrects each 
reading as taken, by adding or subtracting the proper amount. 
In this way all errors due to pressure may be eliminated, so long 
as the gradient remains constant, but an inspection of Plate I, or 
of any barogram, will show that a gradient is not to be trusted 
for any great length of time. Sometimes it will run along for 
hours, or even days, with very little change; this is especially 
true in anticyclonic areas, and again several abrupt changes may 
occur within a few hours. On this account the observer must 
not only test his gradient by stopping at intervals to note the 
rate of change, but he must be continually on the alert for the 
signs which presage a change, such as variation in the character 
or amount of cloud, or the direction or velocity of the wind, etc. 
At first the recognition of those changes which mean a variation 
in gradient will be difficult, but a little practice will make it 
almost instinctive. This method has the advantage that it requires 
the wide-awake, intelligent judgment of the observer all the 
time he is in the field, and, when combined with careful descrip- 
tions and sketches made from all prominent points, yields very 
creditable results. If only one method for the correction of 
these errors can be used, this is decidedly the better, both for 
the reason indicated, and because there are times when the baro- 
metric variation is not the same over even so small an area as 
that indicated. It is desirable however, where possible, to com- 
bine the two, using the first only as a general check on the 
second. 
In spite of all precautions, however, it will be found that there 
are times when the barometer is worse than useless. Such a time 
is indicated on Plate I, beginning at g in the morning of the 25th 
and ending shortly after noon on the 26th, another beginning at 
3 in the afternoon of the 22d and ending at 4 in the morning of 
the 23d. Short disturbances of this kind occur also at 2 A.M. of 
the 21st, and shortly after noon of the 14th. * These’ irregular 
movements almost always occur during storm periods accom- 
panied by gusty winds, and are generally of short duration, though 
I have known them to extend over the greater part of a week. 
