28 Prof. G. H. Darwin, [Nov. 22, 



Fig. 8. 



The ink is propagated along the convolutions of the stem of the 

 ink tree, bunt the convolutions are themselves propagated upwards, 

 and each convolution corresponds to one oscillation. The motion of 

 the ink along the convolutions soon becomes slow, but the convolu- 

 tions become broader and closer. Thus the upper part of the tree is 

 often seen to be most delicately shaded by a series of nearly equidistant 

 black lines. A perfectly normal ink tree, made by a very thin stream 

 of ink, would be like the fig. 9, in which the whole is formed by a 



Fig. 9. 



single line ; but it is not possible to represent the extreme closeness of 

 the lines adequately. 



In the transition from the mushroom stage to the tree stage it 

 appeared to me that it was very frequent that only half the ink tree 

 was formed. At any rate I have frequently noted the mushrooms 

 and half the tree vortices lasting during many oscillations, and then 

 the other halves of the trees gradually appeared. This might, of 

 course, be due to an accidental deficiency of ink in an invisible tree 

 vortex, but I have observed this appearance frequently when there is 

 ink at the stem of the tree, and when there seemed no reason why it 

 should only be carried up in one ascending stream and not in the 

 other. 



If the agitation is very gentle the sand on the crests of the ripple- 

 marks is just moved to and fro ; with slightly more amplitude, the 

 dance is larger, and particles or visible objects, such as minute air- 

 bubbles, in the furrows, also dance, but with less amplitude than those 

 on the crests. When the rocking is gentle the oscillation in the 



