526 Notes on the Siigar Beet and its Juice [July 



where weight is in grams and sugar in percents. Thus, for a dif- 

 ference of 50 grams in weight (the ränge of the weight classes 

 employed by these Bohemian investigators), one would expect an 

 increase in sugar content of only 0.000796 X 50 = 0.00398 per cent. 

 The regression straight Hne and the empirical means are repre- 

 sented in Plate 8.* We note that the observed positive correlation 

 only means a difference of 0.44 per cent. between the lowest and 

 the highest weight grade. 



Data from which correlation surfaces might be prepared are not 

 given for the remaining five series.'^ It would have been possible to 

 get the sign of the remaining five correlations from the data of 

 "Tabelle I and Tabelle VI" by throwing the above formula for 

 r into the form 



S(njwsJ)fN — WS 



Wl 



w $ 



where n is the number of individuals in any weight grade, w, Sw is 

 the mean sugar content associated with this given grade, and .S" 

 denotes a summation for all weight classes. But unfortunately, 

 these two tables are not consistent ! Thus, series 2 ranges from 525 

 to 1475 S^- (mid-ordinates) in one case and from 575 to 1525 in the 

 other. Series 3 ranges from 525 to 1375 gm. in one table and from 

 575 to 1525 in the other. Series 4 begins at 625 and ends at 1375 in 

 one table while in the other it ranges from 725 to 1425. Series 5 and 

 6 are equally faulty in this regard ; in addition, the number of beets 

 for series 5 is given as 173 in one table and as 1390 in the other; 

 the number of beets for table 6 is given as 426 in one table and as 

 173 in the other. Even such confusion as this might possibly be 

 straightened out with some probability of certainty, had not the 



* The two Upper weight classes, connected by broken lines, contain each 

 only a Single beet, so are of little significance. 



5 The data for the first series are given in ponderous detail. The first table 

 gives sugar in tehths of percents for weight groups of 50 gr. ränge. Tables 

 showing correlation between weight and sugar (in groups of o.S per cent ränge) 

 and between sugar and weight are extracted from this as Tables II b and IV. 

 Both of these tables, in essentials identical, are reduced to percentage f requencies 

 and published as Tables III and V. It is really a great pity that the four or 

 five pages thus needlessly used could not have been devoted to the actual data 

 for the five other series upon which their conclusions might have been critically 

 tested. 



