28
SEED TESTING INTERNATIONAL   www.seedtest.org
RULES DEVELOPMENT •
In India: Tests were all on top of paper using an 
8-h light / 16-h dark regime. The tests were set up 
during the light period.
2. The Petri dishes were then placed into plastic 
bags in order to prevent water loss during the test 
and held at 20 ±1 °C. Stacking the Petri dishes 
on top of each other was avoided. If stacking 
could not be avoided, it was limited to a stack of 
two dishes. Care was taken to ensure that the 
temperature was accurate, and the Petri dishes 
were not crowded in the incubator.
3. RE for each replicate was recorded after 72 h 
±15 min. The criterion of RE was ≥2 mm radicle 
protrusion. The data was recorded on data sheets 
provided.
4. After 12 days, for each replicate, a final 
germination count of normal and abnormal 
seedlings was recorded (ISTA, 2023), as well as 
dead seeds.
Information received from all laboratories 
confirmed that the test was completed following 
the guidelines. The only differences between 
laboratories were in light regime and, for one 
laboratory, germination paper, as noted above.
The data were analysed using R function 
ISTAgermMV (2023) developed by the ISTA 
Statistics Committee. Analysis was completed 
for:
• The seven seed lots tested by the five Indian 
laboratories (i.e. three seed lots from Nunhems 
plus four lots from IndoAmerican Hybrid Seeds)
• The three seed lots from Nunhems tested by 
both the European and Indian laboratories (11 
laboratories in total)
 
Only the data for the seven seed lots tested by 
five laboratories in India are included in this 
abbreviated report.
Results
Analysis of Data for Seven Seed Lots 
Tested by Five Laboratories 
The mean laboratory germination for each lot in 
all laboratories was above 85% for all lots except 
lot 4, which had a mean laboratory germination 
of 51% (data not presented).
RE counts after 72 h at 20 °C revealed differences 
between seed lots in each laboratory (Table 3). In 
comparing the overall means for seven seed lots 
in each laboratory, it was revealed that laboratory 
3 had the lowest results for RE (Table 3), 
suggesting that radicles equal to or greater than  
2 mm may not have been clearly identified.
The seed lot mean from all laboratories identified 
two seed lots (1 and 6) with the highest RE 
value and hence the highest vigour, with lot 4 
having the lowest vigour (Table 3). The ranking 
of seed lots was similar within each of the five 
laboratories (Fig. 1). The low RE that suggested 
the lowest vigour for lot 4 was, however, largely 
attributable to its low final normal germination. 
The greater variability in the RE data for lots 7 
and 4 (Fig. 1) is what would be expected for seed 
lots with low vigour.
Tolerances: Considering four replicates of  
50 seeds, five results out of 35 were out of 
tolerance, four of which were from laboratory  
3 (lots 3, 4, 5 and 6) and one from laboratory 2  
(lot 4).
Mixed model for comparing test mean seed lot 
values: Results from fitting the mixed model 
are displayed in Table 4, indicating a significant 
effect of seed lots on RE results.
Repeatability and reproducibility: The value for 
repeatability (⨍𝑟 = 1.06, Table 5) is close to 1.0 and 
is therefore acceptable. The value of 2.09 for 
reproducibility (⨍𝑅 , Table 6) is close to that theo-
retically derived using the over-dispersion factor 
between laboratories established by Miles (1963).
Lab
Lot 1
Lot 2
Lot 3
Lot 4
Lot 5
Lot 6
Lot 7
Lab 
mean
1
80
56
64
50
66
87
54
65
2
88
93
82
38
78
84
84
78
3
58
68
62
30
30
61
30
48
4
89
86
90
74
82
90
82
85
5
86
83
79
48
58
79
56
70
Lot mean 
80
77
75
48 
63
80 
61 
69
Table 3. Mean radicle emergence values for seven seed lots tested in five laboratories; each value is 
the mean of four replicates of 50 seeds
Figure 1. Ranking of the seven seed lots of onion (4 × 50 seeds)
Source of variation
Sum of squares
Mean square
Num DF
Den DF
F value
Pr (>F)
Lot
2478.16
413.03
6
24
8.58
0.000049
Table 4. Results from fitting the mixed model
Table 5: Repeatability estimates for seven lots of Allium cepa using a linear mixed model
(repeatability 
standard 
deviation)
(reproducibility 
standard 
deviation)
69
13.43
8.63
6.94
1.06
16.33
𝑝𝑝̅… 
𝜎𝜎$"#$ 
𝜎𝜎$"%&×"#$ 
𝑆𝑆( 
𝑓𝑓( 
𝑆𝑆)  

View this content as a flipbook by clicking here.