H. M. Parkhurst—Tails of Comets. 39 
agreeing with the computations, for two which are entirely in- 
consistent with the remainder of the series. Perhaps I should 
add that these observations, and indeed all the observations but 
one, were made without any expectation that they would ever 
In predicting the width of a comet’s tail, it may be assumed 
to bear a certain ratio to its length. Up to the time of the dis- 
appearance of the nucleus, the ratio of one-sixth was sufficiently 
exact. The tail was then approaching us, and on this account 
would be expected to appear to grow wider as well as longer. 
Assuming that it remained unchanged in form, it should have 
been about six times as wide on July 21 as on July 13. But 
newt at being 18° wide, it was seen to be only ‘43°, possi- 
y 5°. 
I have refrained from alluding to an important point with re- 
gard to the front edge. When the tail is coming directly 
toward us, the front edge becomes the medial line. en the 
tail is 4° wide, assuming the right hand edge to be the front 
edge would involve an error of 2°; or, if the tail were 18° 
wide, as computed, it would involve an error of 9°. ere is 
no such error; therefore, there is no visible portion of the tail 
to the right hand of the true front edge. If the tail is flat, lying 
in the plane of its orbit, we may readily understand why it 
should be so foreshortened when directed toward us. Indeed, 
this proves too much; for if it were as thin as Saturn’s ring, it 
should have appeared as a mere line instead of being 5° wide. 
If the form of the section were elliptical, it would account for 
the observed width on the 21st of J uly ; but when the tail was 
first seen, the earth was fully 11° above its plane, as seen from 
the comet, and the tail would have appeared much narrower in 
all the observations before July 18. Phe dark line behind the 
nucleus cannot indicate a hollow conical tail, for a diminution 
of the central light by ,':th part would hardly be eo 
but that dark line, and the dark lines separating the different 
or are perfectly consistent with the theory of a thin flat 
il. 
There is another important fact. The front edge was 
toward the right until July 20; but at noon of July 21 the 
earth passed through the plane of the orbit. On the evening 
of J Wy 21, therefore, when the tail was 5° wide, the front edge 
shoul have been either at the left hand, or at any rate at least 
2° from the right hand edge; and yet the latter agreed with 
the line computed from the same formule with those of all the 
preceding days; and although the tail was seen for five days in 
Succession, during which the earth passed through the plane of 
the orbit, there is no apparent discrepancy between the computed 
line and the observed position of the right hand edge of the tail. 
