50 THE STUDY OF PLANT COMMUNITIES * Chapter III 



statistician is able to express mathematically how good his sam- 

 pling may be. Such a method frequently brings several sampling 

 areas into close proximity at the same time that wide areas are left 

 unsampled. Within these wide areas, there are very likely to occur 

 a number of infrequent or unusual species in small numbers, which 

 would be of little concern in a statistical treatment but whose 

 presence could be of great interest to the ecologist. For him, it is 

 usually desirable to have as many of the variations as possible rep- 

 resented in his data because they are subject to interpretation in 

 terms of experience and the nature of related communities. For 

 such purposes, statistical methods are often of little help. It is, 

 therefore, probable that quadrats distributed systematically 

 throughout the stand as evenly and widely as possible are quite 

 satisfactory for most ecological sampling. In fact, systematic sam- 

 pling is likely to be better than random sampling for certain eco- 

 logical purposes. 



Any method that will insure wide and even distribution of sam- 

 ples should be satisfactory. The limits and extent of the stand must 

 first be ascertained, and sampling plans made accordingly. Once 

 the plan is made, it should be followed rigidly unless some previ- 

 ously unknown irregularity, like a swamp or an outcrop of rock, 

 should fall within a sample. 



In small stands it is possible to plan a grid pattern and to sample 

 at regular intervals in this pattern. When stands are large but of 

 reasonable uniformity, it is common practice to run one or more 

 lines across the greatest extent and to space the quadrats evenly 

 along these lines. It would appear that the more widely the plots 

 are spaced in an area to be sampled the greater the efficiency of the 

 sampling unit, provided the spacing is not so great as to make 

 correlation negligible between adjacent plots. 194 Under some con- 

 ditions, it may be desirable to run the lines with a surveyor's 

 transit, although a compass line will usually suffice, and in open 

 country it is possible to run them by sighting on some landmark. 

 The spacing may sometimes require accurate measurement, but 

 pacing may serve quite satisfactorily. The important thing is to 

 avoid any method bordering on personal judgment in placing the 

 plots once the sampling is under way. This should be remembered 

 particularly when the sampling is being done to prove or disprove 



