614 The Philippine Journal of Science 
If these hypotheses are true, type A should always be ac- 
companied by abnormally small refractions; type B, by abnor- 
mally large. 
But No. 42 is explainable by neither hypothesis. It was ob- 
served from high level, 1512 meters, the disk was considerably 
distorted in its descent, there was no material flattening of 
the disk at lower contact or before, yet the end was a thin 
blue line. On account of the difficulty of judging the moment of 
lower contact with a somewhat hazy sea horizon 75 miles away 
I supposed, till I had computed the duration from the records, 
that lower contact had been noted too late. Yet the duration 
came out longer by 21.9 seconds than that computed. The dis- 
tortion was such as to prevent an assignment to either type. 
Diffraction at upper contact has also suggested itself as a 
plausible cause for a real, not merely an illusory, prolongation 
of sunset. It could never show itself with a mirage, but a 
normal or an intensified normal gradient would be favorable. It 
would cause an apparent increase of refraction at upper contact 
as compared with lower. 
