YET A TEAR LATER. 85 



The totality will last longest in North Australia, 

 -where its duration will be more than four minutes, or 

 nearly two minutes longer than the duration of the 

 eclipse of last year at the best stations. In Java the 

 totality will last more than three minutes. In Ceylon 

 the duration of totality will barely exceed by a few 

 seconds the duration of totality last December. A 

 somewhat curious mistake was made on this point in a 

 scientific journal. Mr. Hind, in his first and compara- 

 tively rough estimate of the course traversed by the 

 moon's shadow, had placed Trincomalee on the border 

 of the track, so that the duration of totality at Trin- 

 comalee would have been very short. But after his 

 final and more elaborate calculation, he set Trincomalee 

 close to the centre of the shadow-track, with a duration 

 of total obscuration amounting to two and a half 

 minutes. Strangely enough the increase of the esti- 

 mated duration was alone noticed by the writer of an 

 article in Nature, and it was reasoned that since the 

 duration is so considerable at Trincomalee on the 

 border of the track, it must be very much greater at 

 places on the centre of the track. I need scarcely 

 point out that this inference was unwarranted. In 

 fact, the duration of totality can never under any cir- 

 cumstances be considerable for places close to the 

 border of the shadow-track. 1 In southern India the 

 eclipse will last about as long as the eclipse of last 

 year at the best stations. 



1 A somewhat similar mistake occurred in 1870 whereby the Sicilian 

 eclipse party formed too sanguine expectations of the duration of totality 

 jn that island. 



