152 Diagrams to Illustrate the Effects of the Weather 



The horizontal line in the middle of the diagram is taken as a 

 base line from which the variations from the means in the times 

 of flowering are calculated. If a plant is observed to flower (say) 

 ten days earlier than the calculated date, a shaded line is drawn 

 downwards a certain space, if it is ten days later a black line of the 

 same length is drawn upwards. If several plants have tbeir mean 

 dates of flowering on the same day, the average of the differences 

 from the mean is taken to indicate the state of vegetation on that 

 day. Thus if there are four plants whose mean date of flowering 

 occurs on any particular day, and one was observed ten days later, 

 the second five days later, the third two days earlier, and the fourth 

 five days earlier, then 10 + 5 — 2 — 5=8, and as there were four 

 plants observed, 2 days late would be the average indicated in the 

 diagram. 



There are certain objections (and no doubt serious ones) to this 

 method of registering, which may be remedied hereafter as the 

 number of years of observations increases. The most serious is that 

 the averages are not taken from an equal number of plants for each 

 day. It may be argued that a single species for each day would 

 give better results ; but the objection to this is that a pasture-loving 

 plant may be taken one day, a water-plant another, a wood-plant 

 a third, and so on, and the results would still be very far from the 

 truth. Where all are taken the general state of vegetation is, I 

 conceive, more truly represented. 



Another objection, which I hope to correct in time, is that all the 

 plants have not been observed for the same number of years ; thus 

 the date for one plant may have been computed from seventeen 

 observations, and another from only fifteen, and so on. This error 

 has been partially obviated by only taking those plants which have 

 been observed not less than fifteen times, except in certain very 

 special cases. Twenty years is by no means too short a time for 

 securing proper averages, considering all the errors and omissions 

 which must necessarily occur in such an investigation as the present. 

 Such as it is it can be considered only approximate, and as such I 

 must beg my readers to consider it. 



One other remark must be made. Each year was worked out from 



