A STATISTICAL STUDY IN CANCEE DEATH-RATES. 
By G. D. MAYNARD, F.R.O.S.E., Pretoria. 
As there is still much that remains obscure in the aetiology of Cancer, any 
new light that can be thrown on this important subject seems worthy of careful 
consideration. 
It is with a sense of great diffidence that I bring these figures forward, knowing 
full well how liable to error death-rates may be, and how easy it is to draw false 
conclusions from the study of figures. But when all is said and done it is to 
figures that the final appeal must be made, and if the death-returns are so 
unreliable as to be worthless, then it is certain that there are no more reliable data 
to be obtained. The comparatively few cases that come within any individual 
experience are, from the point of view of exact knowledge, of little or no value. 
If therefore we are to learn anything from the history of the past, it is to 
figures that we must go, guardedly and with great caution, bringing all the 
collateral evidence that we can obtain, to check and control our results, and above 
all using accurate methods of analysis before drawing any conclusions. 
Theories are frequently being advanced, in medical literature, in conflict with 
the records, these either being ignored, or set aside with the remark that they are 
so unreliable that no attention should be paid to them. Marked divergencies 
in the rates are dismissed as being due to errors in registration, the increasing 
cancer death-rate is regarded as being due to better diagnosis, and so on. 
Before entering on any discussion as to the causes of the varying cancer death- 
rates as observed in different countries — and even in the cities and towns of the 
same country — it will be advisable to make a few general remarks on how the data 
in this paper have been obtained and dealt with, and to answer a few of the 
objections that may be raised as to the value of the returns themselves. 
At the outset one must admit that many imperfections are inherent in all 
statistics of death which are not based on post-mortem examinations. But because 
the figures are not absolutely correct are we to neglect them, or regard them as 
absolutely useless ? One must remember that where large masses of figures are 
being dealt with the errors in excess tend to balance those in defect, and wrong 
