Ill] 



OF STATISTICAL METHODS 



91 



explains, and draws deductions from, the resulting series and arrays 

 of numbers. It deals with simple and measurable effects, due to 

 complex and often unknown causes ; and when experiment is not at 

 hand to disentangle these causes, statistical methods may still do 

 something to elucidate them. 



Now as to the growth of man, if the child be some 20 inches, or 

 say 50 cm., tall at birth, and the man some six feet, or 180 cm., 

 high at twenty, we may say that his average rate of growth had 



10 13 



Time in years 



20 



Fig. 4. Curve of growth in man. From Quet^let's Belgian data. 

 The curve H* is proportional to the height cubed. 



been (180 - 50)/20 cm., or 6-5 cm. per annum. But we well know 

 that this is but a rough preliminary statement, and that growth 

 was surely quick during some and slow during other of those twenty 

 years; we must learn not only the result of growth but the course 

 of growth ; we must study it in its continuity. This we do, in the 

 first instance, by the method of coordinates, plotting magnitude 

 against time. We measure time along a certain axis (x), and the 

 magnitude in question along a coordinate axis (y); a succession of 

 points defines the magnitudes reached at corresponding epochs, and 

 these points constitute a ''curve of growth'' when we join them 

 together. 



