20 percent of the total fisherman hours . This 

 is at least partly caused by the inclusion of 

 botli Memorial Day and the Fourth of July. How- 

 ever, the fifth day averages for fish per hour 

 agree fairly well in most instances with the 

 averages based on tlie total samples. Agree- 

 ment is not extremely good for pool 6 and pool 

 10, but even here the error is only about 12 

 percent . 



Table 27 gives for most of the individual 

 pools the every-Sunday averages as compared 

 with the total sample averages. In each instance 

 the result is like that for the total of all pools 

 in that the every-Sunday average is substantially 

 lower than the total average . 



Tables 28 and 29 give information similar 

 to the above for the winter of 1944-45. Several 

 of the pools and localities are given individually 

 in table 28, with a comparison of the averages 

 for every fifth-day with the total averages. The 

 starting day of the fifth-day series was selected 

 by lot. For the most part there is a reasonably 

 good agreement. There are a few exceptions, 

 however, for which it is hard to account. A 

 particular one of these is pool 10. Apparently 

 this discrepancy in pool 10 came about because 

 the fifth-day series by chance included several 

 days when there were large numbers of fisher- 

 men on Gremore Lake and the fishing was poor. 

 This is demonstrated by the figures in table 28 

 for Gremore Lake. Another individual locality 

 for which the agreement was not very good is 

 Bartlett Lake (in pool 6). This lake perhaps 

 contributed to the discrepancy shown for pool 6. 



Table 29 gives for the winter of 1944-45 

 the average fishing success for certain of the 

 pools on Sundays as compared to that on all days. 

 As in the summer fishing, the Sunday fishing is 

 definitely less productive. Pool 10, for some 

 unknown reason, provides an exception. 



Table 30 gives for 3 pools the compara- 

 tive averages for a fifth-day series and for the 

 total for the winter of 1945-46. Here the agree- 

 ment is remarkably good. 



Possibilities for reducing the size of the sample . 

 The effort, and therefore the cost of an exten- 

 sive creel census operation such as this one 



could be reduced materially if the necessary 

 information could be obtained by a smaller 

 sample or a subsample. Whether or not it is 

 possible to do this depends first of all upon what 

 information is desired. The problem is simpli- 

 fied if it can be assumed that the only essential 

 information is that dealing with averages, or to 

 be specific, with the average fishing success 

 for a given locality or area for a given season. 



The (iiscussion above leads to the pre- 

 sumption that for this area of the Mississippi 

 River it is possible to obtain a reasonably ac- 

 curate figure for average fishing success for a 

 season by using some sort of a reduced and 

 randomized sample without the necessity of ob- 

 taining complete coverage for every place and 

 for every day of the season. Some discussion 

 is made here of that point since it may be worth 

 considering in setting up future creel census 

 programs. However, it must be pointed out 

 that these remarks apply specifically to a cer- 

 tain section of the Upper Mississippi River and 

 to the years 1944-46. The extent to which they 

 may be applied to some other time and place is 

 a matter for the judgment of the researcher. 



In the first place, even the large and ex- 

 tensive volume of data supplied by the creel 

 census here reported upon, is in itself only a 

 sample , since as stated, only a part (perhaps 

 30 or 40 percent) of the total fishermen during 

 a given season were interviewed. For the pur- 

 pose of all of the computations herein contained 

 it has been assumed that this sample was 

 random . 



Although it has been shown that for the 

 years involved the fishing success factor was 

 not greatly different for two summer seasons 

 and for two winter seasons, there is no justifica- 

 tion whatsoever for assuming that the fishing 

 success cannot and does not vary between any 

 two years. It is a matter almost of certainty 

 that fishing was very much better in the Upper 

 Mississippi River in the years 1948-50 than it 

 was in 1944-46. Therefore, it is not possible 

 to write an absolute all-time average figure for 

 the fishing success of this body of water on the 

 basis of the average for any one season. U it 

 is desired to know what the average for a long 

 period of years may be, it seems that it is 



51 



