contention in is fullest claims were true definite improbabilities would arise in connection with a number of sets of observations on record.
Mirages are only slightly treated in the usual meteorological text books so a reference to a full treatment and discussion of "cold" class mirages should be useful. This is given in two papers by Fujwhara and Hideka in the Geophysical Magazine of the Central Meteorological Observatory, Tokyo Vol IV, No 4. The problem is treated quantitatively and the diagrams, ray tracing and discussion are most instructive. The mirages there described "Sinkora" are more complex than those dealt with by Vollprecht, and Japanese authors consider the isopycnic surfaces which cause the refraction are not planes but a system of curved surfaces. The mirages appear over the bay on certain occasions when the air is colder than the water but cooler off-shore than near land, a situation which would produce flat isopycnic surfaces at sea but elliptical shaped near shore.
It would not seem that this shape of density surface could be present in most directions at Port Hedland when the phenomena described by Vollprecht were observed and probably this accounts for the simpler appearance. His account and discussion is most interesting and deals with as matter still requiring further treatment. Meteorologists who may desire to take a closer interest in this subject would find a study of Japanese articles helpful.
Forecasting of exceptionally heavy hailstorms is impracticable at present, and, as this article shows, even the forecasting of the occurrence of thunderstorms on the day in question would not have been undertaken with a strong likelihood of success. Adiabatic diagrams constructed from the upper air temperature and humidity recordings give very good indications of probable thunderstorms, but, unfortunately, these records are taken during the night and not near the time of thunderstorm formation. It is therefore necessary, while retaining as most probable the upper level conditions revealed by the earlier ascents, to estimate lower level conditions (such as surface maximum temperature etc.) later in the day about the time of thunderstorm formation. The author in this paper shows that this procedure would have lead