THE TRANSCONTINENTAL ARC. 
219 
acquired we have learned much from experience, and future 
results will throw stronger light on the problems before us. 
What we want, above all things, is a means of measuring the 
force of gravity at sea. 
One word more in general touching sources of inaccuracy 
and the agreement of results. In the work around the Dis¬ 
trict of Columbia it became necessary to observe the Capitol 
dome from both sides and at different times. In the results 
there appeared a discrepancy so large and of such a peculiar 
nature as to call forth an investigation. It seemed as though 
the Goddess of Liberty changed her position from time to 
time. When the results were carefully examined it was 
noticed that in the forenoon she was apparently too far to 
the west and in the afternoon too far to the east. Then it 
occurred to the investigator that possibly this peculiar move¬ 
ment might in some way be connected with the sun. This 
was tested by calculating how much the iron dome might 
be expected to yield by expansion from the influence of the 
sun’s rays, and it was shown that the fair goddess in making 
her daily bow to the ruler of the solar system moved her 
head by an amount quite sufficient to explain the discordant 
observations. 
This case is cited not because it is interesting in itself, but 
for the reason that we have here an example of how simple a 
cause may be whose effects remain long unexplained. There 
are many similar instances. Take the gravity measures. 
Observations with different instruments would not agree 
until some one noticed that the support of the pendulum 
was set in motion by the experiment, and that the air was 
viscous, and therefore adhered to the vibrating body and was 
drawn along after it. The viscosity correction was then in¬ 
troduced as a feature of gravity work. Recently a large dis¬ 
crepancy was found to exist in the direction of a line; the 
cause, now apparent, but for some time unsuspected, is lateral 
refraction, resulting from the proximity of the line of sight 
to the mountain range; and so we might go on noting in¬ 
stances of the same kind. Results that are systematically 
