AGRICULTURAL STATISTICS. 335 



ators, Agents and Correspondents; (4) Uniformity with other 

 countries. 



The System. 



What system should we adopt? We must keep pace with 

 other countries, and yet we must not lose sight of the fact that 

 we are young and have our own_conditions and peculiarities to 

 consider. In other words, we shall have to turn our attention 

 to older countries and base our system on their experience. But 

 here caution is required, and the attempt must not be made to 

 force the coat of the giant on to the shoulders of the young man. 

 The system must be adapted to the requirements of South 

 Africa. 



There are the two systems, the Descriptive and the Numeri- 

 cal. The question arises, which of the two are we to adopt? In 

 my humble opinion I think that an amalgamation of the two will 

 meet our case. Our country is sparsely populated, and does not 

 present any great difficulty in the collection of numerical data. 

 It is the system which we have applied in the Transvaal, so far 

 with success. 



First, we should have estimates (which must be monthly) of 

 the coming crops, and after the harvest the actual yield must 

 be ascertained. The actual yield is a check on the 

 estimates and supplies a basis for the Agents on which to 

 work the following season. 



For monthly or frequent and prompt forecasts the descrip- 

 tive method must be applied. 



There is an idea that the Census could be taken as a base on 

 which to work. The Census is taken every five years. If this 

 be done, then I feel convinced that at the end of each quinquennial 

 period we shall always find ourselves wide of the mark ; and such 

 Statistics are misleading and do more harm than good. 



I would certainly counsel that we begin in a small way, i.e., 

 with the principal products only, especially those that are being 

 exported. Make a good job of it and thus win the confidence of 

 the public and especially of the commercial world. As time rolls 

 on and experience is gained we could gradually expand. 



The machinery ought to be perfect and not beggarly ; but 

 for the first few years we can safely let alone such systems as 

 Logarithmic curves or Algebraic treatment. 



The mistakes of other countries ought to be an object lesson 

 to us, and we must try to avoid falling into similar errors. If 

 we turn our attention to the system in vogue in the United States 

 of America, we find to our disappointment that it is not perfect, 

 to say the least of it ; and I say this, not with any idea of criti- 

 cism, but as a warning. I will not presume to express an opinion 

 but will let better men than myself speak. 



If we turn our attention to the report of the proceedings of 

 a meeting of the Royal Statistical Society, held on May 16th, 

 191 1, we find the following opinions expressed : — 



