Repeatability of subjective grading in fur animals
111. Grading of live blue foxes in different environmental conditions

Hilkka Kenttämies and Kerstin Smeds

Kenttämies, H. & Smeds, K. 1992. Repeatability of subjective grading in fur
animals. 111. Grading of live blue foxes in different environmental conditions.
Agric. Sei. Finl. 1: 315-322. (Univ. Helsinki, Dept. Anim. Breed., SF-00710 Helsinki,
Finland.)

Association between repeatedly scored body size and fur characteristics were studied
in live blue foxes. Gradings in cages and outside cages in lamplight and daylight were
also compared. Colour tended to be easier and clarity more difficult to evaluate than
the other traits. Differences between judges in accuracy of grading were greater than
between various grading environments. The grading was more reliable outside cages
than within cages. The most uniform results were obtained when the same judge
graded the animals in the same environmental conditions.

Key words: fur characteristics, repeated evaluation, cage grading, corridor grading

Introduction

Reliability of subjective grading among fur bearing
animals is highly dependent on the animal species,
the skill of the judge and the environmental
circumstances (Jonsson 1971, Reiten 1977 b,
Jezewska and Maciejowski 1982, Kenttämies
and Käyhkö 1992, Kenttämies and Smeds 1992).
In practice, numerous farmers perform the grading
of live animals themselves, at least in grouping the
animals for breeding or pelting. The grading inside
cages is commonly used as the first rough evalua-
tion in selecting animals for pelting. A more detail-
ed grading of the animals for breeding is often
carried out outside cages by an adviser.

In previous studies with live minks and foxes,
fairly low heritability estimates have been found
for some subjectively scored traits, such as the
quality traits (Jonsson 1971, Reiten 1977 a.
Kenttämies 1988, Lagerkvist and Lundeheim

1990) and clarity (Rosberg and Olausson 1978,
Kenttämies 1988). This may be due to the fact that
the additive genetic proportion of the total variation
for the traits is low or that the data used for the esti-
mation of heritabilities have been less suitable for
these purposes. One reason may be that the traits
have been uncertainly determined or they have
been difficult to judge reliably. The traits are, how-
ever, used for the evaluation and selection ofbreed-
ing animals.

This paper is one in a series concerning the ac-
curacy of grading in fur bearing animals. The reli-
ability of grading for live foxes was discussed in
Part I (Kenttämies and Käyhkö 1992), and that
for mink and blue fox pelts in Part II (Kenttämies
and Smeds 1992). The purpose of the present study
was to ascertain the reliability of grading for
repeatedly scored traits in live blue foxes. The aim
was also to compare evaluations in different grad-
ing conditions conducted by different judges.

315

Agric. Sei. Fint. 1 (1992)

https://www.c-info.fi/en/info/?token=d3Ny9WegfVeWFEEc.7gNwKP7TSsEHGAjFMbZNGQ.SIbaNo0WRa2XLNeto0qlBiIzEPEJ1xLlUMKyQ0JNyrWp9LyZn7JnjZh39nXOsosMtp5t_rZTjxRVpGcLAwmpThGoE3ZtBXUwsE2LG8RWJM6gZn7HM89cZjas81Vo6H4e_TWPtVc_chxbbovvmqZGFOnyjUxhCATsCR6bmhclyvDVGNwANJxKnrfQFaJOx4GSjnhxN5ILsJmMGpStBIJmJMuLyOLie9fHspKcrbUBeoX7_QqPJqTWzPQXTXTGDJ1P5RPNWLqtWMwZPZnWIRfYxxQ4hWvdqKlOIJiZI-W7VQ


Material and methods

An experiment for repeated grading in different
environmental conditions, was carried out at the
Agricultural Research Centre, the Fur Farming
Research Station.

In November 1989, a total of 77 live blue foxes
(35 males and 42 females) were independently
judged by three trained persons, two advisers from
the Finnish Fur Animal Breeders’ Association and
a research technician from the Research Station.
Each judge evaluated the blue foxes three times in
succession in cages in daylight and four times in a
corridor where the judging was done two times in
daylight and two times in lamplight. The last grad-
ing, outside cages in lamplight, was postponed to
the third grading day. The procedure for evaluation
is presented as follows:

Times of
judging

Date of Grading
judging condition
22.11. In cages, daylight 3
23.11. Outside cages, daylight 2
23.11. Outside cages, lamplight 1
24.11. Outside cages, lamplight 1

The animals (with a few exceptions) were graded
a total of 21 times. The following five traits were
evaluated in each grading: body size, colour, density
of underfur, density of guard hairs, and clarity. A
scale from 1 to 5 was used where 1 denotes smallest,
darkest or poorest, and 5 largest, lightest or best.

Weather

The temperature, humidity and lightness in the shel-
ter house were measured at least three times a day.
The temperature varied from -1 to -5°C, while the
humidity remained at 100 % during the experiment.
The quality of light varied in the course of the trial.
For the first day the sun was shining, the second and
third day were cloudy and occassionally snow or wet
snow was falling. On the first day at the start of
grading, the intensity of the daylight was 170 lux. At
noon a value of5000 lux was measured on the sunny

side and 650 lux on the shaded side of the house, in
the afternoon (at 2 pm) 600 lux was found on the
sunny side and 150 lux on the shaded side of the
house. On the second and third day, the intensity of
daylight was 50 lux in the morning, 130 lux at noon
and 20 lux in the afternoon while an intensity of
240-320 lux was measured when a lamp was used.

Statistical methods

The material was edited and analysed using a stat-
istical program WSYS (Vilva 1989). Factors
affecting the five traits were studied using the anal-
ysis of variance with the following model:
Model 1:
Y .. p + a + b. + c.+ d.+ e. ..ijklm = r i j k 1 ijklm

where Y..., = individual observationijklm

(i = general mean
a= effect of the /th judge; /=l-3
h. = effect of the /lh grading condition; j=1-3 (l=in
cages in daylight, 2=outside cages in daylight,
3=outside cages in lamplight)
ck

= effect of the 7th time of judging; £=l-3 (in
cages), £=l,2 (outside cages both in daylight and
lamplight)
d[ = effect of the /th sex of animal; 1=1,2 (1—
females, 2=males)
e = random error

ijklm
The effects of the judge, grading condition, time

ofgrading and sex of animal were considered to be
fixed. Interactions between these factors were also
studied. Data were also analysed separately by
judges and grading conditions. The effect of light
intensity and the interactions between judge and the
intensity of light on the traits were studied within
various grading conditions. The data were also ana-
lysed separately for males and females.

Accuracy of grading was defined as a repeat-
ability coefficient for the repeated scorings. The
repeatabilities were analyzed using random and
mixed models. In the random model the effects of
judge and grading environment were ignored, and
the data were analyzed using the following model:
Model 2:
Y =ii+f + eno " •'n no

316

Agric. Sei. Fint. 1 (1992)



where f = random effect of the nth animal withall
•'ll

other factors as described in Model I.
The foregoing model was also used in analyzing

repeatabilities by judges and grading conditions.
Repeatabilities within judges and grading condi-

tions were analysed according to the following
model:
Model 3:
Y.. =u+ a. + h.. +f. + e..ijno ” i ij J ijn ijno
where A,=random effect ofthe/th condition within
/th judge
f. = random effect of the nth animal within /th
J ijn

judge and jih condition with all other factors as
described in Model 1.

Repeatabilities were analysed separately within
grading conditions according to the following
model:
Model 4a:
Y =ix +a +f +e.mo ” i J m mo
wherefm

- random effect of the nth animal within
/th judge with all other effects as described in
Model 1.

In order to analyse repeatabilities separately
within judges, another model, Model 4b was
applied. The fixed effects of the judge were substi-
tutedby grading condition.

Results and discussion

Means and standard deviations

The means of the traits (Table 1) were near the
median of the scale, 3, while the extreme scores (1
and 5) were seldom used. Clarity was evaluated

Table 1. Number of observations (N), means and standard
deviations (SD) afor scores ofbody size and the fur traits in
live blue foxes.

Trait N Mean SD

Body size, females 873 2.78 0.76
Body size, males 730 3.39 0.87
Colour 1604 3.02 0.72
Underfur density 1604 2.89 0.83
Guard hair density 1604 3.23 0.82
Clarity 1604 3.40 0.71

according to a higher and more narrow scale than
the other fur traits. There existed less variation in
clarity and colour (SD=O.7l-0.72) as compared
with the other traits (SD=O.76-0.87). Large varia-
tions in the traits may be due to the fact that some
of the animals were on a restricted feeding regime
which resulted in changes in appearance.

Factors affecting the traits

The judge was found to have the greatest effect on
the grading of the traits. Differences between
judges appeared to be highly significant for each
trait. Significant differences between grading envir-
onments were found for the traits, apart from body
size (Table 2). The colour of the same animals tend-
ed to look lighter in lamplight than in daylight. The
cleanest colours were found in the corridor in
daylight. In the cage grading the animals were
given higher scores for density ofunderfur as com-
pared with the grading in the corridor. A reverse
situation was found for the density ofguard hairs.

There were statistically significant interactions
between judge and grading condition in colour,
density of underfur and density of guard hairs and
some judgescould not totally follow the same scale
in each environment.

The scores given on different times of grading
appeared to be much the same. Nevertheless, when
the latter evaluation outside cages in lamplight was
postponed to the following day, differences
between the two assessments for body size as well
as interactions between judge and time of grading
for colour, underfur density and guard hair density
tended to be significant.

In cloudy weather, slight differences observed in
daylight intensity did not seem to affect the traits.
In sunnyweather, great differences which appeared
in the intensity ofdaylight were, however, found to
significantly affect the scores for colour, underfur
density and guard hair density. Besides these traits,
significant interactions between judge and daylight
intensity appeared for body size. Nevertheless day-
light intensity was not found to affect the grading of
clarity.

317

Agric. Sei. Finl. 1 (1992)



Table 2. Effects ofjudge and grading condition on scores for body size and the fur traits in live blue foxes.

Factor N Least Squares Means
Body size Colour Underfur Guard hair Clarity

density density

Judge
1 528 3.12 3.02 3.07 3.02 3.72

2 537 2.67 3.41 2.44 2.92 3.23
3 538 3.46 2.72 3.21 3.83 3.29

*** *** *** *** ***

Condition
1 691 3.07 2.92 2.99 3.15 3.33

2 450 3.10 3.02 2.83 3.30 3.50
3 462 3.08 3.20 2.90 3.31 3.41

Ng *** *� *** ***

Condition: 1 ■ in cages in daylight, 2 = outside cages in daylight, 3 = outside cages in lamplight
*** P<o.ool, ** P<o.ol, NS =not significant

In all the data as well as in the analyses done by
judges, males tended to obtain significantly higher
points for body size and density of underfur as
compared to females.

Repeatability of grading

Average associations between therepeated evalua-
tions are given in Table 3. Higher repeatabilities
were obtined after eliminating the differences and
possible interactions between environmental condi-
tions and judges (Within judge, condition and an-
imal) as compared with those obtained from the
animal components only (Within animal). Due to
the obvious effects of judge and environment on
the traits, the previously mentioned repeatability
coefficients are considered more reliable than the
uncorrected ones.

Colour was found easier to judge than the other
traits. On the other hand, clarity and density of
guard hairs seemed to be considerable difficult to
grade. These results are in accordance with previ-
ous reports for minks (Jonsson 1971, Reiten 1977
b, Jezewska and Maciejowski 1982, Kenttämies
and Smeds 1992), for blue foxes (Jezewska and
Maciejowski 1982, Kenttämies and Smeds 1992),

and for silver foxes (Jezewska and Maciejowski

1982, Kenttämies and Käyhkö 1992).

Differences between environments

The highest repeatabilities were mostly obtained
when the grading was conducted in the corridor in
daylight, and the lowest ones in cages (Table 4).
Similar repeatabilities were, however, obtained in
cages and in the corridor in lamplight for colour
and underfur density, the most reliable repeat-
abilities being in the corridor in daylight (r = 0.63,
0.64 vs. 0.76 for colour; r = 0.51, 0.48 vs. 0.71 for

Table 3. The repeatability coefficients ± standard errors for
scores of body size and the fur traits in live blue foxes.

Trait Repeatabilities ± standard errors

Within animal Within judge,
condition and animal

Body size, females 0.39± 0.06 0.58± 0.03
Body size, males 0.34± 0.06 0.59± 0.03
Colour 0.48± 0.04 0.67± 0.02
Underfur density 0.35+ 0.04 0.56± 0.02
Guard hair density 0.26± 0.04 0.56± 0,02
Clarity 0.16±0.03 0.53+0.03

318

Agric. Sei. Finl. 1 (1992)



underfur density). The cage grading produced
clearly lower repeatabilities for clarity than the cor-
ridor grading both in lamplight and daylight
(r = 0.46 vs. 0.60 and 0.59).

In the cage grading, a lower coefficient ofrepeat-
ability for colour was found in the present study
(r = 0.63) than in a previous one for silver foxes
(r = 0.74) (Kenttämies and Käyhkö 1992). How-
ever, similar coefficients were found for body size
(r = 0.56 for blue foxes vs. 0.55 for silver foxes).

The grading of blue foxes in the corridor in day-
light, produced higher repeatabilities in the present
study than in a previous study where two different
groups of judges with varying skills performed the
grading (Jezewska and Maciejowski 1982). In
each study fairly high repeatabilities were obtained
for colour (r = 0.76 vs. 0.61 and 0.67 in Jezewska
and Maciejowski 1982). In the present study a high
coefficient was obtined also for the density of
underfur (r = 0.71), while lower and varying ones
were found by Jezewska and Maciejowski 1982
(r= 0.27 and 0.51).

The use of a lamp was expected to increase the
reliability of grading, higher repeatabilities were,
however, obtained in daylight. Equal repeatabilities
were found only in the grading for clarity. Fairly
low accuracy received in lamplight may be partly
due to the fact that the second judging was post-
poned to the following day. This was found to
change the scale ofgrading which could at the same
time negatively affect the correlations. In addition,
the judges had not become accustomed to evaluate
animals using a lamp.

The evaluations in cages and in the corridor in
daylight were repeated on the same day. During the
corridor grading in daylight, enough light was
available due to a clear covering ofsnow, in spite of
the cloudy weather. In these conditions trained
judges succeeded in correcting for slight changes in
lighting. No previous studies concerning the com-
parison of live animal grading in different environ-
ments seem to have been undertaken. However,
Jonsson 1971 and Reiten 1977 b found obvious
differences in correlations between repeated grad-

Table 4. The repeatability coefficients ± standard errors for scores ofbody size and the fur traits in live blue foxes analysed
by grading conditions.

Trait

Body size, females
Body size, males
Colour
Underfur density
Guard hair density
Clarity

Repeatabilities ± standard errors

Cages Corridor

daylight daylight lamplight

0.451 0.05 0,72± 0.04 0.641 0.05
0.47+ 0.06 0.73± 0.05 0.65± 0.06
0,63± 0.03 0.76± 0.03 0.64± 0.04
0.51+0.04 0.71± 0.03 0.4810.05
0.4510.04 0.7010.03 0.5610.04
0.4610.04 0.5910.04 0.6010.04

Table 5. The repeatability coefficients 1 standard errors for scores ofbody size and the fur traits in live blue foxes analysed
by judges.

Trait Repeatabilities ± standard errors

Judge 1 Judge 2 Judge 3

Body size, females 0.69± 0.04 0.48+ 0.06 0.42± 0.06
Body size, males 0,60± 0.06 0.60± 0.06 0.56± 0.06
Colour 0.68+ 0.03 0.68± 0.03 0.65± 0.03
Underfur density 0.53+ 0.04 0.64± 0.04 0.53± 0.04
Guard hair density 0.61±0.04 0.59±0.04 0.47±0.05
Clarity 0.67± 0.03 0.30± 0.05 0.20± 0.05

319

Agric. Sei. Fint. 1 (1992)



Agric. Sei. Fin!. 1 (1992)

Table
6.

The

repeatability
coefficients
±

standarderrors
forscores
of

body
size
and
the
fur
traits
in
live
blue
foxes

analysed
by

judges
in

different
environments.

Trait

Repeatabilities
1

standarderrors

Cages,
daylight

Corridor,
daylight

Corridor,
lamplight

Judge
1

Judge
2

Judge
3

Judge
1

Judge
2

Judge
3

Judge
1

Judge
2

Judge
3

Body
size,

females

0.57±0.08
0.30±0.10
0.24+0.10
0.75±
0.07

0.7110.08
0.61±0.10
0.82±0.05
0.46+0.12
0.46±0.12

Body
size,
males

0.47±0.10
0.38+0.11
0.50+0.10
0.74±
0.08

0.7710.07
0.69±0.09
0.70+0.09
0.78±
0.07

0.53±
0.12

Colour

0.5710.06
0.7210.05
0.6010.06
0.82+0.04
0.76+0.05
0.72+0.05

0.7710.05
0.5210.08
0.6410.07

Underfur
density

0.5210.06
0.6010.06
0.4410.07
0.64+0.07
0.73+0.05
0.75+0.05

0.4910.09
0.5510.08
0.41+0.10

Guard
hair

density

0.5510.06
0.4110.07
0.3210.07
0.7110.06
0.7210.05
0.68+0.06
0.6910.06
0.6910.06

0.3310.10

Clarity

0.61+0.06
0.2710.07
0.1710.07
0.6910.06
0.3710.10
0.3410.10

0.7510.05
0.3210.10
0.2110.11

ings for the corresponding traits among the same
live and pelted minks.

Differences between judges

Differences between judges in the repeatability
coefficients existed for the most traits (Table 5).
The greatest differences between judges were
found in clarity, the repeatabilities were from 0.20
to 0.67. A previous knowledge of the animals
appeared to be of advantage, as the highest coeffi-
cients of repeatability for most of the traits were
received by the judge working at the experimental
farm.

The repeatability coefficients attained by judges
in different environments varied from 0.17 to 0.83
(Table 6). The difficulty involved in subjective
evaluation is characterized by the fact that only sel-
dom did the same judgeproduce the highest coeffi-
cients in each environment. By far the most uni-
form repeatabilities were obtained in the corridor in
daylight, while the largest differences between
judgesappeared when using a lamp.

Using the cage grading, similar differences
between judges in their ability to grade for colour
were found as previously determined among silver
foxes (Kenttämies and Käyhkö 1992). The
repeatabilities obtained in the present study tended,
however, to be lower than those found for silver
foxes (r = 0.57 to 0.72 for blue foxes vs. 0.68 to
0.82 for silver foxes). Smaller differences between
judges were found for body size (r = 0.48 to 0.66
for blue foxes vs. 0.51 to 0.59 for silver foxes).
Also in the grading outside cages, Jezewska and
Maciejowski (1982) found remarkable differences
between different people in their competence of
grading for total evaluation in blue foxes and
minks, lower differences were found for silver
foxes.

Lower repeatabilities for colour and clarity were
obtained in the present study as compared with an
earlier study for blue fox pelts graded by the same
judges (Kenttämies and Smeds 1992). In live an-
imals graded outside cages using a lamp, the coef-
ficients of repeatability for colour varied from 0.52

320



to 0.64, and for clarity from 0.21 to 0.32. In pelts,
the respective coefficients were 0.77 to 0.90 for
colour and 0.54 to 0.66 for clarity. Therefore it can
be stated that not even a trained judge is capable of
evaluating live blue fox fur as reliably as theirpelts.

Conclusions

A series of experiments investigating the accuracy
of subjective grading was conducted on fur bearing
animals. Repeatability ofgrading for live foxes was
discussed in Part I (Kenttämies and Käyhkö

1992), for mink and blue fox pelts in Part II (Kent-
tämies and Smeds 1992) and for live blue foxes in
different grading conditions in the present study
(Part 111).

In scoring living and pelted blue foxes, the same
judges obtained higher repeatabilities for the res-
pective traits in pelts as compared with live an-
imals. Therefore it can be concluded that evaluat-
ing live fur animals is not as reliable as evaluating
their pelts. It seems that live silver foxes were
easier to judge than blue foxes. On the other hand
blue fox pelts seemed to be easier to evaluate than
minks. In addition, there was a tendency for more
reliable grading among some colour types as com-

pared with some others.
In each paper differences were found between

judges in the scales used for judging. In addition,
differences between the repeatabilities obtained by
various persons were obvious. The accuracy of
grading was greatly affected by the grading envir-
onment, grading in the corridor being more accu-
rate than in cages. However, the first assessment of
appearance can be done in cages in order to group
the animals for breeding or pelting.

In each paper colourwas found to be a fairly easy
trait to evaluate, while clarity tended to be one of
the most difficult ones. Scoring for individual qual-
ity traits, ie. density of guard hairs, and particularly
density of underfur seemed to be easier than for
overall fur quality. Subjective evaluation for body
size seemed to be fairly easy, although a higher
accuracy was obtained by measuring the length of
pelt. Grading for pelt defects appeared to be fairly
difficult.

Acknowledgements.
Pekka Petraan and Simo Paavola, Finnish Fur Breeders’
Association and Pekka Toikkonen, the Research Fur Farm of
the Agricultural Research Centre are acknowledged for care-
fully grading the animals. Financial support was received
from the Technology Development Centre (TEKES).

References

Jezewska, G. & Maciejowski, J. 1982. Repeatability of
exterior trait estimations in different species of fur
animals. 33rd EAAP, Leningrad. 9 p.

Jonsson, M B. 1971. Variasjonsårsaker for noen produks-
jonsegenskaper hos mink. Meld. Norg. Landbr.hogsk.
50, no. 6. 47 p.

Kenttämies, H. 1988. Heritability ofbody size and furqual-
ity of foxes Vulpes vulpes. In: Murphy, B.D. & Flunter,
D.B, (eds.). Biology, Pathology and and Genetics ofFur
Bearing Animals. (Proc. IV Int. Congr. Fur Anim.
Breed. Toronto, Canada, p. 548-556.
& Käyhkö, H. 1992.Repeatability ofsubjective grading
in fur animals. I. Grading of live silver foxes. Agric. Sci.
Finl. 1: 303-307.
& Smeds, K. 1992. Repeatability ofsubjective grading in
fur animals. 11. Grading of mink and blue fox pelts.

Agric. Sei. Finl. 1: 309-314.
Lagerkvist, G. & Lundeheim, L. 1990. Fur quality traits in

Standard mink-Price relationships, heritabi 1ities and
genetic and phenotypic correlations. Acta Agric. Scand.
40: 367-376.

Rosberg, S.O. & Olausson, A. 1978. Nedärvning av vikt
och päls-egenskaper hos standard och safirmink. Vära
Pälsdjur 49: 173-177.

Reiten, J. 1977 a. Arvbarhetsestimater for storrelse og pel-
segenskaper hos nwk mink. Meld, Norg. Landbr.hogsk.
56, no. 16, 12p.
1977 b. Korrelasjoner mellom gjentatte målinger og

vurderinger av pels-egenskaper hos mork mink. Meld.
Norg. Landbr.hogsk. 56, no. 14, 15p.

Vievä, V. 1989. Statistical microcomputer program WSYS.
Univ. Helsinki, Dept. Anim. Breed.

321

Agric. Sei. Fin!. 1 (1992)



Manuscript received February 1992

Hilkka Kenttämies
Kerstin Smeds
University ofHelsinki
Department of Animal Breeding
SF-00710 Helsinki, Finland

Kerstin Smeds
Present adress:
Finnish Fur Breeders’ Association
Box 5, SF-06101 Vantaa, Finland

SELOSTUS

Turkiseläinten subjektiivisen arvostelun toistettavuus
111. Elävien sinikettujen arvostelu erilaisissa ympäristöoloissa

Hilkka Kenttämies jaKerstin Smeds
Helsingin yliopisto

Sinikettujen subjektiivisen arvostelun toistettavuutta tutkit-
tiin Maatalouden tutkimuskeskuksen Turkiseläintuotannon
tutkimusasemalla Kannuksessa. Kokeessa verrattiin arvoste-
lun luotettavuutta, kun eläimet olivat arvostelun ajan häkissä
tai ne nostettiin käytävälle. Häkissä eläimet arvosteltiin päi-
vänvalossa, käytävällä ne arvosteltiin sekä päivänvalossa
että lampunvalossa. Vaijotalon lämpötila, kosteus ja valon
voimakkuus mitattiin. Kolme arvosteluun tottunutta henki-
löä arvosteli 77 sinikettua itsenäisesti 2-3 kertaa mainituissa
olosuhteissa. Eläimiltä arvosteltiin koko, väri, massakkuus.

peittävyys ja värin puhtaus. Värin tummuuden arvostelu oli
helpompaa ja värin puhtauden vaikeampaa kuin muiden
ominaisuuksien arvostelu. Arvostelijoiden väliset erot olivat
suurempia kuin eri olosuhteissa tapahtuvan arvostelun väli-
set erot. Arvostelu käytävällä oli luotettavampaa kuin
häkissä. Suuret valon voimakkuuden vaihtelut häiritsivät
värin ja laatuominaisuuksien arvostelua. Yhdenmukaisim-
mat tulokset saavutettiin,kun sama henkilö arvosteli eläimet
samoissa olosuhteissa.

322

Agric. Sei. Finl. 1 (1992)