SAMPLE PLOTS IN SILVICULTURAL RESEARCH 5 



qualifications, two demand special emphasis: Knowledge of the 

 region and knowledge of statistics. 



Unless the investigator has a thorough knowledge of the region or 

 type, he cannot determine whether his samples are truly represen- 

 tative. The necessary acquaintance with the region can be gained 

 only through well-directed personal experience in the woods. Ex- 

 tensive surveys or investigations are the simplest means of gaining 

 an insight into stand characteristics over a wide area. 



He who deals with samples necessarily deals with statistics. For 

 every investigator who works with sample plots, statistical methods 5 

 are one of the essential tools of his trade, needed in every phase of 

 plot work from establishment to final analysis of data. They con- 

 stitute the only means of dealing satisfactorily with questions aris- 

 ing constantly in every part of a silvicultural study, such as whether 

 the plots are comparable, how many measurements should be taken, 

 what degree of reliability can be expected, how much of the varia- 

 tion in the results is due to the method of treatment and how much 

 to other variables, and what is the estimate of error. 



PLAN OF WORK 



Even with a well-thought-out plan, the quantity of work required 

 to collect the basic data of a single silvicultural experiment often 

 seems staggering. Good advance planning, however, makes it pos- 

 sible to obtain many data with a minimum of effort, and in such 

 form as to lend themselves readily to analysis. Statistical analysis 

 of the data will show the degree to which apparent relationships 

 are significant and how far indications can be relied upon. 



First of all, the purpose of the study should be carefully formu- 

 lated. Much time can profitably be devoted to defining the pur- 

 pose in advance. The choice of methods of procedure can then 

 be made more easily, and it will be unlikely that any needed measure- 

 ment will be overlooked or a mass of unnecessary details included. 



Fisher has said that " Nature will best respond to a logical and 

 carefully thought out questionnaire; indeed, if we ask her a single 

 question, she will often refuse to answer until some other topic 

 has been discussed " (26). Furthermore, the plots of a given project, 

 if properly laid out, are likely not only to contribute toward answers 

 to most of the direct questions asked by the experimenter but to 

 yield information on other subjects. Before beginning a silvicultural 

 experiment the investigator should set down categorically the ques- 

 tions to which in his opinion the experiment may be expected to 

 indicate an answer. Questions thus listed define the nature of the 

 data to be collected and often help in deciding upon the details of 

 methods to be followed in collecting them. 



5 Outstanding works on the fundamentals of statistical methods include texts by 

 Yule (82) and Fisher (27). Descriptions of special statistical technics applicable to 

 forestry problems are available in many other works (1, 12, 13, Hi, 22, 29, 30, 32, 33, 

 37, 38, 39, 15, 77, 81). See also Smith,, B. B. the use of punched card tabulating 



EQUIPMENT IN MULTIPLE CORRELATION PROBLEMS. U.S. Dept. Agr., Bur. Agr. Econ. 1923. 



[Mimeographed.] Methods of preparing volume and yield tables are discussed in the 

 appendix (p. 79). As statistical methods are now rather generally established in 

 biological work and their application to forest research is not particularly difficult, it has 

 seemed logical to omit from this publication a specific discussion of biometry as applied 

 to forestry. 



