TRANSACTIONS OF SECTION A. 453 
Australia, and, perhaps, the Indian Ocean, The abnormal actions or conditions 
giving rise to these large and prolonged variations must hence be persistent for long 
periods, and be effective over the whole of that extensive area, and hence cannot 
be inferred with certainty from the examination of the data of one small portion of 
the area atfected—e.g., India. The variations undoubtedly accompany variations in 
the complete atmospheric circulation over the Indo-oceanic area, and the effective 
forces or actions must be such as to influence the whole movement in a similar 
manner in the two monsoons or seasons of inverse conditions in Southern Asia, 
This inference furnishes a very strong reason for the conclusion that the meteoro- 
logy of the whole area similarly affected from 1892 to 1902 should be studied as a 
whole, and not in fragmentary detail by various weather bureaus, and, as at present, 
without any co-ordination of the results of these bureaus. 
The discussion has also indicated that the south-west monsoon current is a 
periodic or intermittent extension of the permanent circulation of the south-east 
trades to the peninsulas of Southern Asia, and also that variations in the strength, 
volume, and direction of movement of the latter affect the extension, volume, 
aqueous vapour contents, and precipitation of the south-west monsoon currents in 
Burma, India, and Abyssinia. This fact further emphasises the necessity for the 
co-ordination and systematisation of the work of observation in the Indo-oceanic 
meteorological province and the continuous and systematic examination and dis- 
cussion of observations for the whole of that area. 
It is, of course, possible that it may be necessary to extend this work to a 
larger area than the Indo-oceanic region, For Sir Norman Lockyer and Dr. 
Lockyer have shown that similar pressure variations to those of Bombay occur 
over a large portion of the Eastern Hemisphere, and variations of opposite 
a (similar to those of Cordova) over a considerable part of the Western Hemi- 
sphere. 
The Indian Meteorological Department, with the sanction of the Government of 
India, is now arranging to collect and tabulate data for the whole area between the 
Central Asian winter anticyclone and the permanent South Indian Ocean anti- 
cyclone, and to utilise the information for the investigation of the causes of the 
large and general variations of rainfall in Burma and India from year to year. 
This extension of its labour is recognised as necessary for the improvement of the 
seasonal forecasts, an important feature of the work of the Department the value 
and importance of which are fully recognised by the Government of India. 
Possibly the practice of the Indian Meteorological Department in the preparation 
and issue of long-period or seasonal forecasts is considered to be not only 
unscientific, but not justified by comparison with facts. Professor Cleveland 
Abbé, in his paper on ‘The Physical Basis of Long-range Weather Forecasts,’ 
expresses his opinion that ‘ we are warranted in saying that during the thirteen 
years (1888-1900) the only real failure has been that of the prediction of the 
monsoon season of 1899, the year of phenomenally great drought in that country.’ 
This opinion is probably more favourable than I should myself give, but it is the 
opinion of an independent meteorologist eminently qualified to give a judgment in 
the matter. 
My own opinion with respect to weather forecasts is that there appears to be 
too strong a desire for absolute accuracy, possibly due to public and newspaper 
criticism. Certainty is not possible in weatber forecasts based on imperfect 
information, and in which the introduction of a single unknown factor in regions 
beyond observation—e.g.,the upper or middle atmosphere—may completely alter the 
course of events. Percentages of success are an inadequate measure of the utility 
of forecasts. To be of real value as estimates of utility they should be calculated 
rather on the information required, and which might be reasonably expected, 
than on that actually given. 
It appears to me that the striving after perfection in short-period forecasts to 
the exclusion of other claims is impeding the extension and progress of meteorology 
in other useful directions. It is absolutely essential that officials preparing or 
utilising forecasts should recognise that every forecast is based on imperfect 
information and experience, and hence that all important forecasts should be 
