438 Proceedings of Royal Society of Ediiiburgh. [sess. 
a dry clean glass tube of the greatest width which is compatible 
with preventing the air entering and the mercury running out. 
An attempt was made to ascertain the size of the drops into which 
a stream of varying velocity divides, by counting them by means 
of an apparatus which I described in 1878,* by which the falling 
drops actuated a j)en which recorded each drop on a revolving 
cylinder. While it was easily proven that the same law holds good 
for drops formed in this manner as for those dropping from a solid, 
the difficulties of manipulation rendered this method of little value 
for accurate work. 
The second method was carefully investigated, and the results 
are fully detailed in the paper referred to, in which it was clearly 
shown that a liquid dropping from itself (sustained in column form 
by walls which it did not wet) gave the same variation of size with 
rate as when dropped from a solid. 
This method absolutely precludes the idea of a portion of the 
drop being torn back by adhesion to the solid, as there is no solid 
to which the mercury could adhere. 
The conclusion arrived at was that the neck of a drop forms a 
tube through which liquid is flowing into the drop, and after a drop 
begins to fall it receives an accession of liquid during the time 
taken to complete the rupture of the neck, this accession varying 
directly as the flow. Besides this, it seemed probable that owing to 
the “ stump ” of the drop following after the falling drop the life- 
time of the neck might be lengthened. 
This led to a conclusion diametrically the opposite to that of 
Guthrie — viz., that a true drop is one of which the rate is infinitely 
slow, and all other drops are greater than a true drop, and not less, 
as Guthrie concluded. 
The weight of a true or normal drop is therefore easily found by 
determining the difference of weight with rate, and reducing the 
rate to zero, which then gives the weight of a drop at the moment 
it begins to part. 
If it be true that the increase in the weight of a drop with rate 
is due to the influx of liquid through the neck while rupture is 
taking place, it is clear that the size of drops will not be accurately 
* Trans. Toy. Soc. Edin.^ vol. xxiii. 697. 
