104 C. E. VAN ORSTRAND 
these columns is therefore of the convex type. In Table II, the re- 
ciprocal gradientstabulated in thelast column of averages are generally 
larger than those of the next adjacent column. This implies a pre- 
dominance of straight lines and concave curves, as might be expected 
from the shallower depths and the more rapid approach of the wells 
to granite on the crests of the structures. 
The results assembled in Table II are quite consistent; even the 
J-group which is abnormal in the other two tables is here in fair agree- 
ment with the other means. The agreement is not a coincidence. 
It is due to the fact, as previously stated, that the weights in the J- 
group are more and more in agreement with theory as the depth- 
temperature curves approach straight lines. 
The arbitrary rejection of a total of 7 fields in each of which there 
are only 2 wells, group C, does not seriously affect the results obtained 
on either the crests or the flanks of the structures. 
REJECTED OBSERVATIONS 
Theoretically rejected observations in all of the tables have been 
indicated by an asterisk. Rejections occur in Oregon, Texas, and 
Wyoming. The Oregon data are based on temperature measurements 
that are sufficiently precise, but on account of the irregularly shaped 
depth-temperature curves, it is difficult to determine a representative 
gradient from them. The one gradient in the entire list that is most 
likely to be incorrect is the one at Albany, Alabama. Here the 
measurements were made in an uncased well which was 12.5 inches 
in diameter at the top and 8 inches in diameter at 2,000 feet. The 
rise in the depth-temperature curve to a depth of 750 feet is almost 
imperceptible. The results suggest a circulation of water either by 
convection or by the movement of water from one bed to another. 
Theoretical calculations do not indicate that this well should be re- 
jected. The other wells that appear in the list of rejections represent 
extreme geological conditions. They can not be rejected on the basis 
of errors of observation. 
INTERPRETATION OF TABLES I, II, AND III 
Excluding from Table I the Albany well and the wells that are 
known to be in proved or prospective oil fields, we have the following 
series of gradients from the first three columns of tabulations: 
764 
