550 
Smith—Stadia Measurements. 
The practical deductions to be drawn from this work are 
two—1st, do not attempt accurate work under conditions 
not represented in the interval determination; and 2d, during 
the midday hours do not take sights which require the lower 
line of sight to pass nearer than four feet from the ground. 
THE CORRECT STATEMENT OP THE PROBLEM. 
Unfortunately for the interests of the stadia method, the ef¬ 
fect of “differential refraction,” 'in producing accumulative er¬ 
rors, and the use of proper methods for preventing the same, do 
not seem to be understood by stadia engineers. For instance, 
one prominent engineer argues that the best time for determin¬ 
ing the interval, is during the middle of the day , for the follow¬ 
ing reasons: 1st, that during the midday, atmospheric re¬ 
fraction is at a minimum; and 2d, that at such times the change 
in refraction is slight. His misunderstanding results from the 
fact that he is considering absolute refraction instead of differ¬ 
ential refraction. The writer has reason for the belief that 
this misunderstanding is a general one, 
It would be just as sensible to give the usual remedies for 
pneumonia to a patient suffering from consumption. The name, 
or diagnosis, is important because it determines the remedy. 
Absolute refraction is at a minimum near midday, as claimed, 
but differential refraction is near a maximum at such times. 
Stadia measurements are not concerned with absolute refraction, 
but instead, with the relative or differential refraction of the 
two lines of sight, constituting a rod reading. However, if an 
engineer determines his interval during a midday hour, as some 
advise, work done during such hours will show a higher degree 
of accuracy than work of other hours, but unless he intends to 
confine all stadia work to midday, the engineer would be invit¬ 
ing large systematic errors by choosing such time for the in¬ 
terval determination. 
That this is not mere theorizing, a glance at the results of the 
recent survey of St. Louis will abundantly prove. The report 
of this survey shows that all distances were read too long, a 
ct which the records certainly prove. Now let us in- 
