180 REPORTS ON THE STATE OF SCIENCE. 



as definitely as possible, and to assign limits to the precision of the 

 estimate of the aggregate income of this group. Every estimate should 

 be criticised in the light of its effect on this aggregate, and if any amend- 

 ment is thought necessary, its effect on the result should be computed 

 before its importance is accepted. As a by-product we have also 

 obtained valuable information as to special classes, but we are bound to 

 make estimates in all cases whatever our ignorance may be. It is 

 believed that the use of the modulus, chosen so as to include all esti- 

 mates which are reasonably probable, enables us to make bricks without 

 straw which will be strong enough for the strain we shall put on them. 



Class 1. Civil Service. — There is no exact classification extant of the 

 individual salaries of Civil Servants, but the numbers and the salaries 

 are given in great detail in the estimates for the Civil Services and 

 Eevenue Departments published annually. The only difficulty is to 

 interpret such items as ' 39 second-division clerks, minimum 70L, annual 

 increment 101., maximum 250L, total 5.400L,' and to estimate from 

 them the number and income of those who have not more than 160L 

 It is believed that this has been done in such a way that the result, 

 both in number and distribution of salaries, is very near the facts, 

 and that it was unnecessary to try to get complete details from the 

 Treasury. The process used is to take every item by itself, and divide 

 the groups according to whatever indications are given. A further 

 difficulty is to divide the wage-earners from the salaried, but the Census 

 Instructions to Tabulating Clerks make it possible to select with very 

 fair precision those who are included under the heading Officers and 

 Clerks. As a result of our examination we find 56,000 male Civil ser- 

 vants (officers and clerks) in the United Kingdom, of whom about 

 19,000 received more than 160Z. The Census total was 52,000, which, 

 when increased by 9 per cent, to bring it up to date, corresponds closely 

 with the total just found. The average salary of these clerks is, 

 approximately, 951. Our conclusion is that there are 36,000 4- 2,000 

 male clerks, with not more than 160L, and that their average is 951. ±51. 

 As regards the women clerks, we only find 14,300, whereas the census 

 gives 18,000. [The difference is not improbably due to many sub-post- 

 mistresses returning themselves as Government servants, whereas, in 

 fact, they would be more properly classified as shopkeepers, and only 

 receive part of their income from the Government. We have trans- 

 ferred these to class 27. Sub-postmasters and mistresses are not 

 properly included as Government servants.] The main bulk of 

 these female clerks are in the Post Office service. Only 175, 

 or about 1 per cent., of these women earn more than 160Z. The 

 estimate we adopt under this heading is 15,000 ± 2,000, at 57Z. ± 51. 

 In this and in other classes a small number pass the exemption limit in 

 virtue of other sources of income. There is no means of allowing for 

 this except by the limits shown by the ' modulus.' 



Class 2. Local Government. — We have received in answer to our 

 Schedule of Inquiry thirty-one detailed returns from county boroughs 

 in England and Wales, with an aggregate population of over 4,000,000,, 

 and a return frp,m the London County Council. There is con: 

 siderable difficulty in definition in this class, and it is improbable 

 that I be town clerks who made returns for us obtained exactly 



