12 THE IMPOVERISHMENT OF THE SEA. 



summary to show which of the averages may be taken as reliable, and 

 which are based on an insufficient series of observations. As the 

 figures ostensibly represent the average conditions prevailing during 

 periods of five years' duration, it is manifestly impossible to regard any 

 of them as satisfactory which are based on the surveys of one or two 

 years only in each period, especially if the years fall exclusively near 

 the middle of the decade. The averages for the second period may 

 be accepted as thoroughly satisfactory, so far as the number of years is 

 concerned ; but in the case of the first period the averages for January, 

 February, March, and possibly December, may justly be discredited, 

 either on account of the insufficiency of the number of years repre- 

 sented by the averages (January — one year only), or by the fact that 

 the two years included are limited to the latter portion of the period 

 (February and March, 1889 and 1890; December, 1888 and 1890). 



Four of the monthly averages out of the twelve are thus eliminated 

 upon merely preliminary examination of the data upon which they are 

 based. Further scrutiny shows that an equally serious objection may 

 be urged against several of the remaining averages, owing to the 

 une<|ual representation of the two areas in the combined averages. 

 The great differences between the Firth of Forth and St. Andrews Bay 

 in regard to the seasonal abundance of the different kinds of flat-fish 

 render it imperative that in any combination of the averages for 

 comparative purposes the two areas should be represented in equal 

 proportions during the two periods. Yet during the second period, 

 while the Forth was investigated with almost perfect regularity month 

 by month during the successive years, there are four months (January, 

 May, August, and September) in which no examinations whatever were 

 made in the Bay for four years out of the five. For these months, 

 therefore, during the second period, the influence of the Forth largely 

 predominates in the " averages "; whereas during the first period the Bay 

 and the Forth were ecjually represented, so far as the number of surveys 

 is concerned, in three out of the fovir months (viz. January, May, and 

 August). On this count, therefore, the January averages are still 

 further discredited, and we are also forced to add May and August to 

 the list of unreliable averages, which brings the total up to six out of 

 the twelve. 



That the fallacy caused by disproportionate representation of the two 

 areas in the two quinquennial periods has led to errors of an appre- 

 ciable and serious character may be judged from the following figures. 

 They represent approximately the average number of fish of the 

 different kinds distinguished taken in one haul of the trawl in each 

 nionth of the year in the closed waters, the numbers for St. Andrews 

 Bay being kept distinct from the numbers for the Firth of Forth. They 



