Adje0011 e PDF
Adje0011 e PDF
ADJE0011-E-PDF
Appendix X5
Distribution Analysis
of Sieve Openings
Copyright 2024
ASTM International
100 Barr Harbor Drive
P.O. Box C700
West Conshohocken, PA
19428-2959
USA
[Link]
ASTM Designation E0011
ASTM Stock Number: ADJE0011-E-PDF
Regarding
Appendix X5. Distribution Analysis of Sieve Openings
The goal of this project was to inspect the openings in samples of sieve cloth from multiple suppliers
using the same vision system to eliminate equipment / technique as a variable in results.
ASTM E29.01.02 selected the number of sieve designations at 11 to best consider a broad range,
specifically U.S. No. 8, 20, 40, 60, 80, 100, 120, 140, 200, 270 & 325.
MSIvision has 30+ years experience in machine vision systems and applications. With an unlimited
budget, it would be possible to use hardware components that could provide superior measurement
results; however, the selected hardware and approach was deemed adequate to achieve the project
goals for normality analysis.
The equipment used for all images was a Keyence XG-H500M 5 Mega Pixel monochromatic camera
and 5X telecentric lens, while for coarse mesh a larger lens was used. The image acquisition and
processing hardware used was a Keyence XG-X2702 system, able to perform high speed image
processing under control of a custom software program.
The inspection involved selecting 100 openings (both warp & shute) from each sample and computing
the width and height of these openings using image processing.
The data is to be analyzed by a third party to determine normality distribution; the inspection results
being in pixels vs. dimensional units should be irrelevant for this analysis.
The inspection results are presented in 22 excel workbooks, as 11 sieve designation samples were
gathered from each of 2 suppliers (simply noted “A” & “B”).
Within each workbook there are 4 worksheets for each sample of the same designation, ex. A-8-S1
thru A-8-S4, B-8-S1 thru B-8-S4, etc.; hence 88 data sets of 100 openings in both the warp and shute
direction.
The data sets are arranged in 25 rows x 8 columns, 4 for Width (or warp opening) and 4 for Height
(or shute opening).
This is due to the inspection finding 25 random images per sample, and recording measurements from
4 opening locations per image.
The following shows a mesh sample on the flat red collimating filter backlight with camera/lens above.
Each piece was marked with the mesh count and identifying orientation of the warp wires.
MSIvison comment upon review of the data is that consideration should be given to
eliminating outlier data prior to analysis due to isolated anomaly results.
- The coarser mesh sometimes shows a number of outliers with a '0' value. This could ultimately be
fixed, as the reason is an imaging issue due to the larger opening size. The search region for edges is
correspondingly larger and, in some cases, may extend outside the image boundary, leading to a 'fail' in
the tool and a 0 result.
- Manual placement of samples in the camera view is not precise, so there may be some slight angular
rotation in some images that can affect the width and height results. A more rigorous fixturing
approach (beyond the scope of the current project) could minimize this effect.
- Some samples were difficult to hold fully flat (mostly the coarser mesh), which these cases can result
in minor data errors.
MSIvison comment regarding the vision system used is that inspection variability
was minimized.
The same image processing parameters were used for consistency which provided acceptable
results for all cases, specifically:
Bin = 0, refers to whether binary thresholding on the image was used, switch on (1) or off (0)
Len = 1.6, refers to the length of the edge search window relative to the opening size
Win = 40, refers to the width of the edge search window relative to the opening size
The image below shows an opening for finding edges; blue rectangles are search regions;
green lines are found edges. The edge tool averages the edge points and creates the green line
representing the best fit of the averaged points.
This particular example from the software development platform indicates the opening height
measurement is 23.173 pixels.
For the purposes of determining if values are normally distributed this is acceptable, but if the goal was
to accurately seek the actual dimension, edge tool averaging becomes an issue and the search area
might be reduced to a single, centered line with averaging omitted.
Aim
The aim of this analysis is to assess whether sieve opening measurements follow a normal distribution.
Data
The data consists of measurements of openings of sieves. Four sieves of each of eleven sieve designa-
tions were selected, from two different sources/manufacturers (labeled A and B), and for each sieve,
width and height measurements were made of 100 openings. Hence there are 88 datasets, each with
100 observations of width measurements and 100 observations of height measurements.
Summary of conclusion
Based on the Shapiro-Wilks test for normality, we find that between 20 to 40% of the 88 sets of sieve
opening measurements are not normal at the 0.01 significance level. After accounting for multiple
testing using a Bonferroni correction, about 11 to 25% of these datasets are not normal.
However, many of these instances involve one extreme outlier (in addition to any zero-valued
measurements). If one extreme outlier were removed in such cases, as little as 2-5% of these data sets
tested significantly non-normal after a Bonferroni correction for multiple testing (8 to 20% before the
correction).
Methods
The Shapiro-Wilks test was used to test each data set of measurements of 100 sieve openings, with
the height and width measurements tested separately. We also compared results with a more general
Kolmogorov-Smirnov test. The Shapiro-Wilks test is considered a more powerful test, i.e. it is more
able to find non-normality in instances where that is the case. We used a significance level of 0.01
(i.e. probability of concluding non-normality when in fact the distribution is normal, is 0.01). Normal
probability plots were also made to visually inspect the distribution of the data relative to normal
distribution values.
Note that, as suggested by the data documentation, zero values in the data set were removed
before testing for normality. The documentation also indicated that the openings may not have been
1
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
aligned exactly and may be slightly rotated. However, without further information on this, we were
unable to modify the analysis to account for it. We expect the impact on the results to be minimal.
Details of analysis
Table 1 shows the results of the Shapiro-Wilks and KS tests, where “TRUE” means that the respective
tests concluded that the corresponding data set was normal. The last line in the table shows the
proportion of the data sets that tested normal. The left side of the table shows results without
accounting for multiple testing, while the columns on the right of the table show results after applying
a Bonferroni correction. As mentioned in the summary, we find that 20 to 40% of the 88 sets of
sieve opening measurements are not normal at the 0.01 significance level, 11-25% after a Bonferroni
correction.
The sieve measurements came from two sources, denoted A and B in the data file names. The
proportions of data sets that tested normal were similar between the two sources (see Table 2).
There were 4 sieves used for each sieve designation. The proportions that tested normal are shown
in Table 3. These values would be more variable (since there are computed out of 4 observations),
but it appears that the sieves with smaller openings tended to tested normal at a higher proportion.
This could be due to the measurement process.
The normal probability plots are shown at the back of this report. As is often the case, non-
normality occurs at the tails of the distributions. For example, for the widths in A-200-S4 (page 11,
5th row, last plot), there are many values in the upper tail of the data distribution that deviates from
normality, and this data set tested significantly as non-normal.
Note, however, that there are many instances where all except one measurement fall within
the shaded banded indicating normality. Examples are, A-60-S1 (widths), A-80-S1 (widths, and
heights), B-60-S4 (widths). This single measurement tends to be an extremely small measurement
(though, in B-20-S3, for example, there was one very large width measurement). If we remove one
measurement (and only one measurement) when the measurement is more than 4 standard deviations
from the mean, the proportion of datasets that tested non-normal (using the Bonferroni-corrected
Shapiro-Wilks test) decreased to 5% and 2% for widths and heights respectively. Of those that still
tested non-normal, a few of them were because there were 2 extreme observations, and still tested
non-normal because only one measurement was removed (e.g. B-325-S1, widths and heights, and
B-325-S3 widths). These results can be found in Table 4. Updated versions of Tables 2 and 3 based
on this new analysis can be found in Tables 5 and 6 respectively.
Hence, a large number of the data sets tested normal, especially if a single outlying observation
is excluded. Conclusions of non-normality were often due to a single extreme measurement, but not
always. Of course, if all extreme observations were removed, every data set would test normal.
A lot depends on what these outlier values mean. These outliers could be due to errors in the
measurement process e.g. measurement of an opening at the edge of an image of the sieve, and not
related to the sieve itself. If this is the case, more understanding of the measurement process and
the resulting measurements could provide guidance for the data processing needed before use. The
outliers could also be due to a sieve having a small number of unusually sized openings (i.e. the
2
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
measurement process was actually accurate). What this means with respect to the quality of such a
sieve is beyond the scope of this report.
3
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
Shapiro KS Shapiro (BC) KS (BC)
SieveWidth Height Width Height Width Height Width Height
B-8-S1
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-8-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-8-S3
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-8-S4
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-20-S1TRUE FALSE TRUE FALSE TRUE FALSE TRUE TRUE
B-20-S2
FALSE FALSE TRUE TRUE FALSE TRUE TRUE TRUE
B-20-S3
FALSE FALSE TRUE FALSE FALSE FALSE TRUE FALSE
B-20-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S1
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S3
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S1TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S3TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S4
FALSE TRUE TRUE TRUE FALSE TRUE TRUE TRUE
B-80-S1TRUE FALSE TRUE TRUE TRUE FALSE TRUE TRUE
B-80-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-80-S3TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-80-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S1
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S2
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S1
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S2
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S3
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
B-120-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S1
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S2
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-200-S1
FALSE TRUE TRUE TRUE FALSE TRUE TRUE TRUE
B-200-S2
FALSE FALSE FALSE TRUE FALSE TRUE FALSE TRUE
B-200-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-200-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S1
FALSE FALSE FALSE FALSE FALSE FALSE FALSE TRUE
B-270-S2
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S3
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-325-S1
FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
B-325-S2
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-325-S3
FALSE TRUE FALSE TRUE FALSE TRUE FALSE TRUE
B-325-S4
FALSE TRUE FALSE TRUE FALSE TRUE FALSE TRUE
0.60 0.81 0.83 0.90 0.75 0.89 0.83 0.92
Table 1: Results of test of normality (Shapiro-Wilks and Kolmogorov-Smirnov tests) on each of
the 88 data sets of measurements of 100 sieve openings. The four columns on the right show results
after a Bonferroni correction for multiple testing. Numbers at the bottom show proportion of data
sets testing normal.
4
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
Shapiro (BC) KS (BC)
Sieve Width Height Width Height
A 0.73 0.91 0.80 0.91
B 0.77 0.86 0.86 0.93
Table 2: Proportion of data sets that tested normal (after a Bonferroni correction), for sieves A
and B separately.
Table 3: Proportion of data sets that tested normal (after a Bonferroni correction), for each sieve
size designation separately.
5
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
Shapiro KS Shapiro (BC) KS (BC)
Sieve Width Height Width Height Width Height Width Height
A-8-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-8-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-8-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-8-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-20-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-20-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-20-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-20-S4 FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-40-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-40-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-40-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-40-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-60-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-60-S2 TRUE FALSE TRUE TRUE TRUE TRUE TRUE TRUE
A-60-S3 TRUE FALSE TRUE TRUE TRUE TRUE TRUE TRUE
A-60-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-80-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-80-S2 FALSE TRUE TRUE TRUE FALSE TRUE TRUE TRUE
A-80-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-80-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-100-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-100-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-100-S3 FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-100-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-120-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-120-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-120-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-120-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-140-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-140-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-140-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-140-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-200-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-200-S2 FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-200-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-200-S4 FALSE TRUE TRUE TRUE FALSE TRUE TRUE TRUE
A-270-S1 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-270-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-270-S3 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-270-S4 FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-325-S1 FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-325-S2 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
A-325-S3 TRUE FALSE TRUE TRUE TRUE TRUE TRUE TRUE
A-325-S4 TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
6
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
Shapiro KS Shapiro (BC) KS (BC)
Sieve Width Height Width Height Width Height Width Height
B-8-S1FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-8-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-8-S3FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-8-S4FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-20-S1TRUE FALSE TRUE FALSE TRUE FALSE TRUE TRUE
B-20-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-20-S3FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-20-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S1TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S3FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-40-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S1TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S3TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-60-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-80-S1TRUE FALSE TRUE TRUE TRUE TRUE TRUE TRUE
B-80-S2TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-80-S3TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-80-S4TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S1
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S2
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-100-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S1
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S2
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-120-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S1
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S2
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-140-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-200-S1
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-200-S2
TRUE FALSE TRUE TRUE TRUE TRUE TRUE TRUE
B-200-S3
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-200-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S1
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S2
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S3
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-270-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-325-S1
FALSE FALSE TRUE FALSE FALSE FALSE TRUE FALSE
B-325-S2
FALSE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
B-325-S3
FALSE TRUE TRUE TRUE FALSE TRUE TRUE TRUE
B-325-S4
TRUE TRUE TRUE TRUE TRUE TRUE TRUE TRUE
0.80 0.92 1.00 0.98 0.95 0.98 1.00 0.99
Table 4: Results of test of normality (Shapiro-Wilks and Kolmogorov-Smirnov tests) on each of
the 88 data sets of measurements of 100 sieve openings, where one extreme observation is removed
(if it exists) The four columns on the right show results after a Bonferroni correction for multiple
testing. Numbers at the bottom show proportion of data sets testing normal.
7
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
Shapiro (BC) KS (BC)
Sieve Width Height Width Height
A 0.95 1.00 1.00 1.00
B 0.95 0.95 1.00 0.98
Table 5: Same as Table 2, but with one extreme observation removed, if it exists.
Table 6: Same as Table 3, but with one extreme observation removed, if it exists.
8
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
A−8−S1 A−8−S2 A−8−S3 A−8−S4
127 14 37 17 48 95
127
127
127
Widths
Widths
Widths
Widths
125
125
125
125
15
123
123
123
123
44 92
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
24 59 64
25
125
125
Heights
Heights
Heights
Heights
120
115
5
1 6 48
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
12 25 72 66 2
46.5
46.5
47.0
46.0
Widths
Widths
Widths
Widths
46.0
45.5
45.0
45.5
42
45.0
7 33
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
86 89 93
48.0
48
45 46 47 48
86
Heights
Heights
Heights
Heights
47
46.5
46
45
45.0
42 29 100 66
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
94 73 43 58 36
23.8
23.8
23.8
24.0
Widths
Widths
Widths
Widths
23.4
23.4
23.5
23.4
23.0
23.0
23.0
23.0
96 7 98
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
60 91 21 3 87
23.6
23.5
64
23.2
Heights
Heights
Heights
Heights
23.0
23.0
22.5
22.6
22.4
79 28
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
9
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
A−60−S1 A−60−S2 A−60−S3 A−60−S4
375
28 45 39 36
375
370
Widths
Widths
Widths
365
340
365
355
360
310
27 30 8 52
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
390
380
14 30 32 2
375
Heights
Heights
Heights
Heights
370
360
365
350
355
71 51 71 49
340
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
275
250
270
270
Widths
Widths
Widths
Widths
150
265
260
50
40
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
25 46
250
270
265
Heights
Heights
Heights
Heights
260
150
260
255
250
50
250
44 26 63 81 66 29
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
10 47 22
232
230
230
Widths
Widths
Widths
Widths
150
228
225
226
220
224
86
222
50
12 41 52 52
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
61 68 63
225
225
220
150
Heights
Heights
Heights
Heights
215
215
210
50
200
205
85 60
205
10 58 34
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
10
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
A−120−S1 A−120−S2 A−120−S3 A−120−S4
29 9 80 58
190
190
188
72
100 150
Widths
Widths
Widths
Widths
186
184
184
50
182
180
178
30 38 30
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
195
18 12 34 16
195
60
195
190
190
Heights
Heights
Heights
Heights
190
185
185
185
185
62
180
175
87 59
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
74 41 40 40 35
163
160
160
160
161
Widths
Widths
Widths
Widths
120
150
159
156
80
140
48 157 71 79
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
35 12 77 18
160
158
156
155
156
Heights
Heights
Heights
Heights
154
152
152
145
91
150
148
148
10 60 85
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
39 89 6 46 50
118
100
112
Widths
Widths
Widths
Widths
112
114
60
110
108
108
20
49 52 13
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
57 72 50 3 100
112
100
35
112
Heights
Heights
Heights
Heights
110
110
110
60
108
108
108
20
13 84
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
11
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
A−270−S1 A−270−S2 A−270−S3 A−270−S4
45 69
70
80
70
Widths
Widths
Widths
Widths
79
79
50
50
78
30
77
30
44
77
9 89 91 93 25
10
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
2 4
81
70
80
80
Heights
Heights
Heights
Heights
79
50
78
76
77
76
30
11
74
72
75
70 93 60 25 25
10
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
70
25 75 66 14 13
70
70
94
68
50
68
68
Widths
Widths
Widths
Widths
66
66
66
30
64
64
64
10
71 49
62
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
70
40 71 18 27 19
70
54
68
68
68
68
Heights
Heights
Heights
Heights
66
66
66
66
64
64
64
62
64
62
3 41
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
12 44 98 12 67 47
128
127
126
126
126
Widths
Widths
Widths
Widths
125
124
124
123
122
122
122
20 55
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
22 65
121 123 125 127
41
126
124
Heights
Heights
Heights
Heights
122
122
120
118
10 25 16 5 88
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
12
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
B−20−S1 B−20−S2 B−20−S3 B−20−S4
47.0
51
8 26 26 56
46.5
49
49
Widths
Widths
Widths
Widths
46.0
45.5
47
47
45.0
45
44.5
45
27 36 90 41
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
10 11
47
50
10
47
40
46
Heights
Heights
Heights
Heights
48
46
45
30
46
36
45
44
20
26 24 10 99
44
43
44
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
6 10
23.0
23.0
23.0
Widths
Widths
Widths
Widths
22.8
22.6
22.6
22.6
22.2
22.4
35 25 22.2 26 17 19 26
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
80 92 87 92
23.4
44
23.2
23.2
Heights
Heights
Heights
Heights
23.0
22.8
22.8
22.6
22.4
22.4
87 73 23
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
12 59
362
362
360
Widths
Widths
Widths
Widths
355
358
358
84
356
73
354
354
345
22 62
352
69 65
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
88 83
366
366
370
Heights
Heights
Heights
Heights
362
362
360
69
358
358
42 52 26 40 10
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
13
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
B−80−S1 B−80−S2 B−80−S3 B−80−S4
54 15 82 34 28
274
275
272
Widths
Widths
Widths
Widths
270
270
268
265
266
264
6 75 45
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
266
18 63 97
21
265
260
262
Heights
Heights
Heights
Heights
260
260
258
240
255
56
250
254
82 29 86
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
21 21 16 41
224
232
232
73 73
Widths
Widths
Widths
Widths
220
226
226
216
220
220
23 17
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
85
230
9
225
225
225
Heights
Heights
Heights
Heights
220
215
215
215
205
205
12 12
210
53 52 38 45
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
71 16 63 60 61 64
175 180 185 190
Widths
Widths
Widths
100
180
50
176
24 14
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
91 27 73
185
100 150
180
Heights
180
Heights
Heights
Heights
180
175
50
170
170
53
170
18 66 24 32
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
14
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
B−140−S1 B−140−S2 B−140−S3 B−140−S4
160
156
Widths
Widths
Widths
Widths
156
154
152
151
73 74
152
86 61
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
154
154
75 71 8 82
153
152
151
152
Heights
Heights
Heights
Heights
151
150
150
149
148
149
148
74 73 83 50
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
15 6 57 57
21 21
113
113
100 105 110
100
Widths
Widths
Widths
Widths
110
110
80
60
107
107
44 80
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
114
79
18
114
112
Heights
Heights
Heights
Heights
110
110
110
108
87 87
106
106
106
104
1 59 86 86
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
98 30 11
85
76
84
38
85
70
80
Widths
Widths
Widths
Widths
80
80
50
75
75
76
70
30
36 70 63
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
84 28 38 74
74 76 78 80 82
82
76 78 80 82
75
Heights
Heights
Heights
Heights
78
65
8
74
55
79 68 4
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
15
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG
B−325−S1 B−325−S2 B−325−S3 B−325−S4
71
55
60
60
26
60
83
69
Widths
Widths
Widths
Widths
40
40
40
67
20
20
20
65
94 15 73 84 15
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
78 63 24
69
68
68
86
50
Heights
Heights
Heights
Heights
67
66
66
30
65
94
64
64
30 96 69
10
−2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2 −2 −1 0 1 2
16
&RS\ULJKWE\$670,QW O DOOULJKWVUHVHUYHG 0RQ-DQ87&
'RZQORDGHGSULQWHGE\
8QLYHUVLGDG7HFQRORJLFDGH3DQDPDSXUVXDQWWR/LFHQVH$JUHHPHQW1RIXUWKHUUHSURGXFWLRQVDXWKRUL]HG