Marcu 28, 1912] 
NATURE 
97 
(2) A second proposal is that the teachers of a 
subject at the four colleges should form a board of 
examiners—either four (or only two) to constitute the 
board. Presumably each member would set a portion 
of the papers; in this case, if the four men act, the | 
students at each college would recognise the “pet"’ 
questions of their teacher, which, although forming 
only a portion of the paper, would receive 
answers than the rest of the paper, and this would 
mean, practically, that each college would be holding 
its own examination. Consequently the result would 
be essentially the same as in the first case. Moreover, 
the suggestion that all four teachers should cooperate 
is not quite so feasible as would appear; they would, 
of course, have to meet on several occasions, and 
fuller | 
though it is easy enough for a man in Edinburgh | 
or Glasgow to run up to London in a few hours to 
confer with his co-examiner, yet the geography of 
New Zealand renders travelling less easy. Auckland 
and Dunedin are separated by nearly 900 miles, and 
this journey occupies at least sixty hours. It would 
be very inconvenient, to say the least, for these two 
men to spare time to meet, even at a midway point, 
while the cost to the University of such a scheme 
would be very heavy. Moreover, details of procedure 
would be far from easy to arrange. 
(3) The purely external system of examination is 
condemned by most authorities. The real feature of 
the grievance lies not so much in having the examina- 
tion for degrees conducted by external examiners in 
Britain or elsewhere, as in the total exclusion of the 
teachers from this examination; and it seems to us 
that the best suggestion is one made by two or three 
of those consulted, viz. that the teacher of a subject 
should make a report on each student, which would 
be forwarded to the examiner, who would take it into 
consideration in his award. For it is manifestly un- 
just to a candidate who has worked well throughout 
the year to be judged only by his answers to a paper, 
written on a day on which he may be unwell or other- 
wise unfit. 
Every student, before presenting himself for the 
degree examination, has at present to pass an exam- 
ination held by his teacher, and in the case of science a 
practical examination in addition must be done to 
his satisfaction. The marks awarded in these, if sent 
to the external examiner, would influence him in his 
award. 
Indeed, it happened on one occasion that the degree 
had to be awarded entirely on these college examina- 
tions, for the ship conveying to England the candi- 
dates’ answers was wrecked, and all the papers lost. 
The reformers cavil at the small encouragement the 
university colleges give to research, while, as the 
pamphlet points out, there is opportunity but for a 
limited amount of original investigation. They 
rightly complain of the bugbear of examination if it 
be regarded as the “be-all ‘and end-all” of university 
training; but, since the examination is part of the 
British system precedent to obtaining a degree, it is 
hopeless for a small colony like New Zealand to 
attempt to eradicate this evil so long as the Mother 
Country adheres to it. 
In New Zealand there is no leisured class who can 
afford to spend time in pursuing knowledge for its 
own sake, and the degree is chiefly required by those 
entering the teaching profession, who must have a 
fairly all-round training in subjects useful for their 
purpose. 
To such men and women specialisation at an early 
stage in the university career would be fatal to their 
prospects; there is no demand for specialists in 
chemistry or physics or biology, and it would be a 
cruel thing to encourage a man to spend two or three 
NO. 2213. VOL. 89] 
years in research, with no available opening at the 
end. Moreover, the libraries and staffing of the 
colleges are insufficient, as the reformers emphasise, 
for extensive research, which is best left to the later 
stages of a man’s career, viz. for honours. What 
| sort of research can a student in New Zealand pursue 
in languages? 
It seems clear, however, that certain reforms are 
needed, but we fear that the reformers must not 
expect that all their grievances will he rectified 
immediately. 
EXPERIMENTAL ERROR IN AGRICUL- 
TURAL INVESTIGATIONS. 
| view of the large number of agricultural experi- 
ments carried out in the country it is very desir- 
able that some attempt should be made to put them 
on a sound basis, so that the results shall have some 
permanent value and admit of definite interpretation. 
The experiments cost a good deal of money, practically 
all of which is found by public bodies, and the work 
is frequently carried out without any particular regard 
to scientific method. 
Perhaps the most serious defect hitherto has been 
the ignoring of experimental errors, so that only in 
very few cases could the experimenter say what degree 
of accuracy he had obtained or what was the me 
nificance of the differences he observed. In order 
provide a remedy a day was devoted to the sabieet 
at the agricultural subsection to the British Associa- 
tion in 1910, and some of the papers then read have 
been amplified, and are now issued as a supplement 
to The Journal of the Board of Agriculture. 
They are all couched in simple language, and bring 
home the fact that the value of an experiment depends 
on the degree of confidence that can be attached to 
the result. The opening paper, by Messrs. Hall and 
Russell, deals with field trials, and the general con- 
clusion is reached that the probable error attaching 
to a single experiment is at least +10 per cent. It is 
possible to reduce the error to about +2 per cent. by 
repeating the experiment simultaneously on a number 
of plots, which need not be more than 1/5o0th acre in 
extent. 
The second paper, by Prof. Wood, discusses 
analytical results, the sampling of crops, field trials, 
and feeding experiments, and contains frequency 
curves and tables of odds, setting out the least sig- 
nificant differences in these usual conditions of the 
various classes of determinations. The agricultural 
experimentalist will do well to submit his figures to 
the simple tests suggested here. 
Mr. Pickering deals with experimental errors in 
horticultural work, which are fairly considerable, and 
commonly ignored. The experiments and their inter- 
pretation are more difficult than in purely agricul- 
tural work, and according to the quantity estimated 
may vary from +16 to +20 per cent. for a single 
tree, or from +6 to +8 per cent. for a set of six 
trees. 
Milk investigations are discussed by Mr. Collins. 
An ordinary fat analysis is shown to be liable to an 
error of +0°03 per cent., while the error in the 
solids-not-fat determination can be reduced to o'05 
per cent., but may be higher. 
The Board of Agriculture has undoubtedly rendered 
very useful service by issuing these papers in so cheap 
a form, and it is to be hoped that they will be used 
as extensively as the importance of the subject war- 
rants. 
1 Surplement No. 7 to the Journal of the Board of Agriculture, zorr. 
