566 Mr. Arthur Stanley Eddington [March 26, 



of rotation ; the transition from direct to retrograde motion is thus 

 a perfectly gradual one ; the satellites of Uranus, in fact, illustrate 

 the intermediate stage ; they may be said to move neither backwards 

 nor forwards but sideways. Briefly speaking, the effect of the sun, 

 and the tides on the planet raised by the sun, would be to tilt over 

 the planet until its axis was in the plane of its orbit— the inter- 

 mediate position I have just mentioned. 



In the case of a very viscous planet the tilting may proceed still 

 farther, until, in some cases, the planet goes almost completely ovei ; 

 that is what seems to have happened in the case of Mars and the 

 Earth. The Moon appears to have split off from the Earth after this 

 somersault had been performed, and to have had no share in helping 

 it. But in the case of the other planets, the satellites have had a 

 large share in driving the plane of the planet's rotation right over. 

 Thus Jupiter was tilted rather more than half way by the action of 

 the solar tides, and then he evolved the inner satellites, and it was 

 by their influence that the inversion was completed. Neptune, on 

 the other hand, being very remote from the sun, began to turn 

 over slowly ; before the process had gone very far, the satellite was 

 evolved, and the plane of rotation began to sink back again to the 

 original position ; thus the motion and dii'ection of rotation remain 

 retrograde. 



As to what happens to the satellites when the planet turns over, it 

 is almost more difficult to make a generalisation. In some cases 

 the circumstances are such that a satellite, which separated from 

 the planet when it was rotating backwards, would follow the planet's 

 equator and turn over with it. Probably this is what happened in 

 the case of lapetus and of J YI and VII. But in other cases, 

 namely, those of Phoebe and J YIII, the satellites are left to retain 

 their original motion. 



We see now why it is only the extreme outermost satellites in 

 these two systems that move backwards. They weie the earliest to 

 split off, and refer to a time before the tilting of the planet by the 

 solar tides had proceeded very far. When the theory was put 

 forward, it was predicted that, should any very distant satellites of 

 Jupiter and Saturn be found, they also should revolve backwards. 

 An opportunity for testing this seemed soon to occur. The sixth and 

 seventh satellites of Jupiter were found far outside those previously 

 known ; when, therefore, it was found that they moved forwards, the 

 hypothesis received something of a set back. However, it has turned 

 out that these two satellites were not distant enough from Jupiter ; 

 the still more remote J VIII moves backwards, and greatly strengthens 

 the theory. 



Comet C 1908. 



We must now leave the system of Jupiter and turn to another 

 subject which has been occupying attention at Greenwich, and, 



