Experiments on the Potato- Crop. 561 



Of "The Queen of Flukes" and "Flour-ball," there were no 

 experiments with 6 oz. sets. 



The relative productiveness of the several varieties grown 

 from 8 oz. sets, planted at intervals of 12 inches, stand thus : — 



tons. cwts. qrs. lbs. 



LateEed 38 19 2 25 



Spencer s Kino; of Flukes 34 2 14 



Queen of Flukes 30 5 2 9 



Flukes .21 9 3 19 



Lapstones 4 17 26 



Early Prolific Kidney 3 15 1 11 



The above four series of comparisons are tolerably uniform, 

 as expressing the relative productiveness of the varieties they 

 include. The actual order of precedence of some of the indi- 

 vidual varieties, that do not differ much in their produce, varies 

 a little ; but the relative positions are in general uniform ; the 

 late red in each set of experiments produced the heaviest crop ; 

 and the Early Prolific Kidney appears in every case at the 

 bottom of the list. 



Of the three varieties of Fluke, the greater productiveness ol 

 both Spencer's King and the Queen of Flukes, than that of the 

 ordinary variety, is very noticeable ; Spencer's King especially, 

 throughout the series, producing from half as much more, to 

 twice as much as the Common Fluke, not only in the general 

 averages, but in all the individual experiments. 



Seventhly. Accidental variations of Result. 



It has been necessary, in drawing our conclusions, to altogether 

 avoid relying on the results of isolated experiments. Whatever 

 precautions may be taken to ensure uniformity in the conditions 

 under which agricultural experiments are conducted, unaccountable 

 anomalies in the result will be found to occur ; variations which 

 affect all agricultural crops, and which should be fully recognised 

 and guarded against when inferences are drawn from experiments. 



The only way to remove such sources of error is to throw 

 together the average results of a number of independent experi- 

 ments, so that the irregularities tending in either direction may 

 neutralise each other. I would cite, by way of illustration, the 

 individual trials making up the average results given under the 

 first head. 



At page 554 it was stated that the average balance on 13 ex- 

 periments, in favour of 2 oz. over 1 oz. sets, was 1 ton 13 cwt. 

 "2 qrs. 7|- lbs. per acre ; but if we come to details, it appears that, 

 out of these 13 experiments, 5 show a result in favour of the 

 1 oz. sets, and 8 in favour of the 2 oz. This proportion, 8 to 5, 

 taken by itself, is not very striking, and might be accidental ; 

 but when the sum of the weights of the gains in favour of the 



