UNIVERSITY OF 
 
 ILLINOIS LIBRARY 
 
 AT URBAN A -CHAMPAIGN 
 
 AGRICULTURE 
 
NQN CIRCULATING 
 > 
 
 CHECK FOR UNBOUND 
 CIRCULATING COPY 
 
*. 
 
 / 
 
 White Hulless Popcorn: 
 
 FIFTEEN GENERATIONS OF SELECTION 
 FOR IMPROVED POPPING EXPANSION 
 
 By B. L. Weaver 
 A. E. Thompson 
 
 Bulletin 616 
 UNIVERSITY OF ILLINOIS AGRICULTURAL EXPERIMENT STATION 
 
CONTENTS 
 
 Review of Literature 4 
 
 Materials and Methods 5 
 
 Experimental Results 6 
 
 Variability within the 15 selected generations . . 7 
 
 Analysis of the relationship of progeny to 
 parental selections and nature of variation 
 during the last eight generations of se- 
 lection 10 
 
 Relationship of popping expansion to supple- 
 mentary ear characteristics 12 
 
 Discussion and Conclusions 13 
 
 Summary 17 
 
 Literature Cited . . 18 
 
 Urbana, Illinois September, 1957 
 
 Publications in the Bulletin series report the results of investigations made 
 or sponsored by the Experiment Station 
 
FIFTEEN GENERATIONS OF SELECTION 
 
 FOR IMPROVED POPPING EXPANSION 
 
 IN WHITE HULLESS POPCORN 
 
 By B. L. WEAVER and A. E. THOMPSON 1 
 
 POPPING EXPANSION the increase in volume of the popcorn 
 kernel after popping is an economically important quality factor. 
 Greater expansion not only is a factor in improving eating quality; it 
 also contributes to higher yields of the popped product. Each volume 
 increase in expansion means an increased return to the commercial 
 popper of about $4.50 from a 100-pound bag, according to Eldredge's 
 calculations (10)*. 
 
 Exploratory tests with commercial varieties of popcorn at the Illi- 
 nois Agricultural Experiment Station were begun in 1937. Popping 
 tests showed wide variation in both quality and popping expansion. In 
 1938 a two-ear sample of a white hulless type was obtained from a 
 farmer in central Illinois who claimed to have grown this strain for 25 
 years. 2 A portion of the seed was planted in 1938 and, when compared 
 with other hulless varieties, it was found to be later in maturity, far 
 more vigorous, and higher yielding. The quality and ability to pop were 
 considered only fair. It was thought that a recurring type of selection 
 for increased popping expansion over a period of years would result in 
 a general improvement in the performance of the strain. The new ac- 
 cession was grown in larger quantity in 1939 and the initial selection for 
 improved popping expansion was made. Fifteen generations of selec- 
 tion were completed in 1953. From this material was developed the 
 high-quality, open-pollinated variety, Illinois Hulless. Numerous inbred 
 lines have also been developed and experimental hybrids produced. 
 
 The objective of this publication is to present the significant findings 
 of this study, and to point out certain implications suggested by the 
 breeding method that was employed. 
 
 1 B. L. WEAVER, Assistant Professor of Vegetable Crops, Emeritus, and A. E. 
 THOMPSON, Assistant Professor of Vegetable Crops. The selection and experi- 
 mental work was initiated and solely conducted by the senior author. The junior 
 author contributed to the statistical analyses and preparation of the manuscript. 
 
 2 Mr. Malcomb Chapman, Farmer City, Illinois. 
 
 * This and similar numbers refer to the literature citations on page 18. 
 
4 BULLETIN No. 616 [September, 
 
 Review of Literature 
 
 Brunson (2) demonstrated that continuous selection of seed ears on 
 the basis of individual popping tests over a seven-year period could 
 improve the popping expansion of a strain. The variety Sunburst, which 
 was subsequently named Supergold, was developed by this method 
 from the unselected variety Queen Golden. The average popping ex- 
 pansion was raised from 19.3 volumes for Queen Golden to 26.1 vol- 
 umes for Sunburst. Brunson stated that the ears of Sunburst were less 
 variable in popping expansion than Queen Golden, and the occasional 
 very poor ear frequently found in unselected sorts had been virtually 
 eliminated. 
 
 Factors affecting popping expansion are adequately discussed and 
 summarized by Huelsen and Bemis (17). Additional information is 
 available on popping expansion and the derivation of varieties and 
 hybrids of popcorn (3, 11, 25). 
 
 Willier and Brunson (28) conducted rather extensive experiments 
 on the relationship of kernel characteristics to popping expansion of 
 White Rice and Yellow Pearl varieties. The kernel characters studied 
 include: percentage of soft starch; weight of 100 kernels; number of 
 kernels in 25 cc.; and length, breadth, and thickness of kernel. The 
 correlations involving expansion indicate that in general those ears 
 with smaller kernels and those with kernels containing the least pro- 
 portion of soft starch gave the greatest increase in volume on popping. 
 They further concluded that large kernels were more likely to have a 
 large proportion of soft starch than small kernels. Length, breadth, and 
 thickness of kernel were all negatively correlated with expansion. 
 Length and thickness were more highly correlated with starchiness than 
 with weight, and breadth was correlated about equally with starchiness 
 and with weight. The multiple correlation coefficient between expansion 
 and the kernel characters was calculated. Only approximately one-third 
 of the variability of expansion is accounted for by variations in the size 
 and starchiness of the kernel. Lyerly (23) concluded from correlation 
 studies that smaller, shorter, and narrower kernels tended to give the 
 highest popping expansion. These data agree with those of Willier and 
 Brunson (28) for weight, length, and width of kernels. Lyerly found, 
 however, that thickness of kernel was positively correlated with pop- 
 ping expansion. Density of kernel and the ratio of thickness to width 
 were also positively associated with popping expansion. 
 
 Crumbaker et al. (7) investigated the inheritance of popping expan- 
 sion by studying segregating progenies of dent-popcorn crosses. They 
 concluded that low popping expansion was partially dominant over 
 
1957] SELECTION FOR IMPROVED POPPING EXPANSION 5 
 
 high. They observed that two backcrosses to the recurrent parent were 
 sufficient to recover popping expansion equal to that of the popcorn 
 parent inbreds. Johnson and Eldredge (21) studied the performance of 
 recovered popcorn lines derived from outcrosses to dent corn. The re- 
 covered popcorn lines were agronomically superior to the original pop- 
 corn parents. Among 74 recovered lines tested in crosses, 31 percent 
 were significantly lower, 7 percent significantly higher, and 62 percent 
 not significantly different in popping expansion from the performance 
 of the original parents. Johnson and Eldredge (21) suggest that since 
 recovery of popping expansion was not difficult, the character must be 
 rather simply inherited. 
 
 Some inbred lines of popcorn have been found to be superior to 
 others in transmitting popping expansion and popcorn-type ears to the 
 progeny, thus indicating differences in combining ability among the 
 lines tested (7). Data presented by Grissom (13) and Clary (4) also 
 demonstrate the existence of different genetic systems for the inherit- 
 ance of popping expansion among the popcorn lines they studied. A 
 heritability value of 70 percent for popping expansion was calculated 
 by Grissom (13). Calculations made from two sets of single-cross data 
 presented by Clary (4) give heritability values of over 90 percent. The 
 minimal estimate of the number of genes controlling popping expansion 
 in Grissom's (13) study was three to four. Clary's (4) estimates of 
 minimum numbers of effective factors among several inbred lines 
 ranged from one to five. Clary concluded that the estimates were too 
 low, but the maximum number might be less than 20. 
 
 Materials and Methods 
 
 The initial selection of 201 ears was made in 1939. These ears and 
 those selected in subsequent years were tested for popping expansion. 
 Only those ears that gave high values for popping expansion were 
 saved for seed to be planted ear-to-row the next year. From 1939 to 
 1945 the number of ears sampled yearly varied from 200 to 320. 
 
 In 1946 the procedure was changed slightly to what might be termed 
 a modified mass selection or "truncation selection," in which seeds for 
 the next year's plantings are taken only from ears having the highest 
 expansion within each of five populations. The large number of indi- 
 vidual ear selections was reduced to those from the top five lines, which 
 were given numbers from 1 to 5. From 1946 to 1953 the experimental 
 procedure remained constant. Each of the five lines was grown in plots 
 containing 14 rows of 16 hills each. The spacing was 3'x3' with three 
 
6 BULLETIN No. 616 [September, 
 
 seeds planted per hill. The five plots were isolated from other corn 
 plots. However, the five individual plots were contiguous and linearly 
 deployed in a north-south direction. The positions of the individual 
 plots were not randomized, but were maintained in the same systematic 
 arrangement throughout the course of the experiment. Sixty ears from 
 each of the five plots were selected for testing each year and seed from 
 the top-ranking ears from each of the five plots was planted the follow- 
 ing year. Usually seed from the top three or four ears within a line was 
 sufficient to make the next year's planting. 
 
 The selected ears were all tested for popping expansion. A 25-cc. 
 sample of shelled corn was conditioned to the optimum moisture con- 
 tent by a method adapted from that developed by Dexter (8). The 
 samples were then dry-popped in an ordinary flat-shaker popper. The 
 popper was fitted with ball-bearing rollers that ran on a small track 
 mounted over a gas burner. The popper was agitated at a constant 
 speed by means of a connecting rod, eccentrically driven by a small 
 electric motor. To reduce variability, check samples from a uniform lot 
 of Illinois Hulless were popped. Every fifth sample was a check, and if 
 any significant variation occurred in the performance of the check, 
 necessary adjustments in popping procedure were made. The volumet- 
 ric increase of the corn after popping was measured in a graduated 
 glass cylinder. 
 
 Additional data were collected on each of the selected ears: ear 
 length, ear width, ear weight, weight of shelled corn, and shelling per- 
 centage. Correlation analyses were made to determine the possible as- 
 sociation of popping expansion and these supplementary measurements. 
 Notes were also taken on whether the kernels were pointed or rounded. 
 
 Experimental Results 
 
 The method of selection utilized proved to be effective in improving 
 the level of popping expansion within the population. The average 
 popping expansion of the population in 1953 was approximately 57 
 percent greater than the average of the original population in 1939 
 (Fig. 1). The means of the populations for the last five generations are 
 actually above the highest variates of the first five generations. The 
 lowest variates of the last two generations are approximately four vol- 
 umes higher than the population means of the first two generations. It 
 would appear that a maximum level had not been reached at the termi- 
 nation of the experiment, although there is some indication that the 
 maximum was being approached. 
 
1957} 
 
 SELECTION FOR IMPROVED POPPING EXPANSION 
 
 44 
 40 
 36 
 
 </, 32 
 til 
 
 3 28 
 
 | 24 
 
 z 
 
 S 20 
 Q 
 
 | 16 
 I 12 
 
 Y* 20.40+ 1.103 (10.I06JX 
 
 - : MEAN 
 
 || = RANGE 
 
 5 6 7 8 9 10 
 GENERATIONS OF SELECTION 
 
 14 15 
 
 Improvement made in popping expansion during 15 generations of selection within 
 open-pollinated, white hulless type popcorn. The graph shows the means and best- 
 fitting, straight-line trend, with the range of highest and lowest variates for each 
 generation. (Fig. 1) 
 
 Variability within the 15 selected generations. The variability 
 within the total population tended to decrease (Fig. 2). Variability 
 within the population has been measured by the standard deviation, 
 coefficient of variation, and Weinberg's formula (27). The Weinberg 
 constant (W) was calculated by the formula: 
 
 s v (Highest variate lowest variate) 
 v (Highest variate mean) (Mean lowest variate) 
 
 where s = the standard deviation of the sample. 
 
 The magnitude of the standard deviation and that of the Weinberg 
 constant tend to be parallel. The Weinberg constant is modified by the 
 skewness of the distribution, which is, to a certain extent, taken into 
 account by the denominator of the formula. The extent to which the 
 distributions of the yearly populations are skewed is clearly indicated 
 in Figure 1 and Table 1. Since the Weinberg constant takes the degree 
 of skewness into account, it should present a more accurate picture of 
 
BULLETIN No. 616 
 
 [September, 
 
 the variability than the standard deviation. The slight downward trend 
 for the standard deviation in part accounts for the increased reduction 
 in the magnitude of the coefficient of variation, since the means for 
 each generation tend to increase. 
 
 The significance of the regression coefficients for the three meas- 
 ures of variability were tested by means of the t test. The t values were 
 2.10, 3.94, and 2.34 with 13 degrees of freedom respectively for the 
 standard deviations, coefficients of variation, and Weinberg constants. 
 The latter two regressions were significantly different from zero at the 
 0.01 and 0.05 levels of probability respectively. The regression coeffi- 
 cient of the standard deviations was not significantly different from 
 zero at the 0.05 level, but did have a probability value between 0.05 and 
 0.10. One may conclude from these tests that variability in popping 
 expansion within the population has definitely decreased over the 15 
 generations in which selection has been practiced. 
 
 24 
 22 
 
 20 
 16 
 
 16 
 ^< 
 
 2 14 
 * 
 S . 12 
 
 10 
 
 ,-Sx2 
 
 Y(WxlOtl9.90-.375(.l60)X 
 Y(C.V.) 20.97-. 888(.225>X 
 Y(sx2)9.3l-.206(.098)X 
 
 5 6 7 8 9 10 
 GENERATIONS OF SELECTION 
 
 12 13 14 15 
 
 Effect of selection for increased popping expansion on the extent of variability 
 within the 15 generations as measured by the Weinberg constant (W), coefficient 
 of variation (C.V.), and standard deviation (s) ; actual data and best-fitting, 
 straight-line trends. (Fig. 2) 
 
1957] 
 
 SELECTION FOR IMPROVED POPPING EXPANSION 
 
 d 
 111 
 
 i o 
 
 JQ'OOmSo-'J'ioO'-oSooooo no 
 
 
 f 
 
 
 a 
 
 ifl t^ (N >-i <O O W 00 M 
 
 | 
 
 
 ts cs cs (N cs ts cs ts cs n ro <*> r> PO ro <N 
 
 
 
 B 
 
 r 
 
 3 
 
 a 
 c 
 3 
 
 
 
 T 
 
 <)>0<5OM 9 
 
 
 
 H 
 
 6 
 
 
 
 Y 
 
 
 
 
 [ 
 
 8 
 
 c 
 
 
 
 4 
 
 00 
 I 
 
 ^Hr*>^ioes ^H 
 
 
 
 "o g 
 
 8| 
 
 
 in 
 
 ^c 
 
 B 
 
 2 
 a. 
 
 c 
 _o 
 
 
 
 Ok 
 
 
 C o) 
 
 S S 
 so 
 
 
 
 
 
 3 
 
 
 
 
 
 5 OOOCSOvvOrO'OOPcS^'OO 
 
 o -^ 
 
 
 
 
 -H ro 10 l NO t^ \0 0> 
 
 g volumes 
 
 4 
 
 s 
 
 OS 
 
 
 
 OOO'^C*^ l ') 0^ Hp O p '5'O r '5 f< 5OOO O 
 
 g 3 
 
 .2 g 
 
 * 
 
 
 d 
 
 H 
 
 1 
 
 c 
 _o 
 
 
 ion of poppir 
 
 
 04 
 
 s 
 
 IIIII^ ! 
 
 s 
 
 C/3 o 
 
 o 13 
 
 d (volumes) 
 
 
 i 
 
 B 
 
 
 ige distribut 
 
 CN 
 CS 
 
 O^OfOesOOsOt^ -f^fjroro OO 
 
 .2 c 
 
 to (0 
 
 I 
 
 i 
 
 M 
 
 d 
 
 y, 
 
 tc 
 O 
 
 du 
 
 III! ! 
 
 Percent 
 
 
 00 
 
 1 
 
 CS CS ^H *-i . 
 
 H 
 'S.J 
 
 
 
 c 
 
 
 
 Ok 
 IO 
 
 7 
 
 II : "I : : : : ] 
 
 fin -0 
 
 "o 
 
 
 d 
 
 X 
 
 I 
 
 c 
 _o 
 
 III! ! 
 
 
 
 OS 
 
 j 
 
 
 1 V 
 
 . o, 
 
 CM O 
 
 
 
 1 
 
 g-^ssi^s s 
 
 
 e 
 
 <o 
 
 OS 
 
 <N 
 
 0) 
 
 3 
 
 
 d 
 
 c 
 D 
 
 2 
 cu 
 
 ir, X ^- r i -^- ir. O ^- 'T 
 
 
 
 
 
 
 
 
 
 eNM-t^ci--N M- 
 
 jaqiunM 
 
 "-i 1 . 
 
 :;":;!!!!!; 
 
 
 
 ar tested 
 
 ^ 
 
 o 10 /i \o oo <- o CN t* 
 
 s 
 
 B 
 i 5 2? 5 S in 10 10 g 
 
 
 
 ^ 
 
 
 
 iS 
 
 
 j 
 
10 BULLETIN No. 616 [September, 
 
 The composition of the population was markedly changed over the 
 15 generations of selection (Table 1). The frequency distributions also 
 indicate the extent of the heterogeneity of the generation variances, 
 which was confirmed by Bartlett's test for homogeneity ( 1 ) . No actual 
 determination can be made to estimate to what extent the heterogeneity 
 is attributable to seasonal fluctuations of the environment or fluctua- 
 tions in the frequency of genetic factors conditioning popping expan- 
 sion. In addition, there may be an interaction of genes and environ- 
 ment, which should not be overlooked as a possible source of variation. 
 
 Analysis of the relationship of progeny to parental selections and 
 nature of variation during the last eight generations of selection. 
 
 A closer analysis of the variation and effectiveness of selection can be 
 obtained by studying the last eight generations of selection. In 1946 
 the top five lines were selected and continued throughout the remainder 
 of the experiment. The five lines actually trace back to only two 
 original ears tested in 1939 as shown below: 
 1939 
 
 71 
 95 
 
 The relationship of progeny to parent is clearly evident if the means 
 for the yearly selections within the five lines and their progenies are 
 studied (Table 2). The correlation coefficient for the association of 
 selection and progeny means was + 0.834, which is highly significant. 
 Approximately 70 percent of the variability in the popping expansion 
 of the progeny can be accounted for by the performance of the parental 
 selections. The regression of the means of the selections on the means 
 of their subsequent progenies was calculated. The regression coefficient, 
 b = 0.678 0.073, was highly significantly different from both and 
 1 as tested by the t test. These data indicate to what extent one might 
 predict the performance of progenies of selections made for improved 
 popping expansion. Over the eight-year period the progenies averaged 
 3.5 volumes lower than the selections from which they were derived. 
 
 An analysis of variance was computed for popping expansion of the 
 tested progenies of selections made within the five open-pollinated lines 
 during the last eight generations (Tables 2 and 3). The analysis is 
 presented with some reservation since the variances from generation 
 to generation were highly heterogeneous, as was previously pointed 
 out. The interaction of generations X lines was highly significant when 
 
1957] SELECTION FOR IMPROVED POPPING EXPANSION 11 
 
 Table 3. Analysis of Variance for Popping Expansion of Five Open- 
 Pollinated Lines Tested and Selected for Eight Generations 
 
 Source of variation 
 
 Degrees of freedom 
 
 Mean square 
 
 
 7 
 
 1 
 6 
 
 1 
 2 
 1 
 1 
 
 1 
 
 3,388.01* 
 19,714.26* 
 666.97* 
 
 255.21* 
 456.61* 
 266.61* 
 481.51** 
 51.71 
 31.03 
 
 45.10** 
 11.45 
 
 
 
 
 
 Lines 
 
 4 
 
 1-1-2+3 vs. 4 + 5 . 
 
 
 Among 1 , 2 and 3 
 
 
 1 vs. 2 +3 
 
 
 2 vs. 3.. . 
 
 
 4 vs. 5 
 
 
 
 28 
 
 Within lines within generations (Sampling error) 
 
 2360 
 
 Total 
 
 2399 
 
 
 
 * Exceeds 0.05 level of significance. 
 ** Exceeds 0.01 level of significance. 
 
 tested against the sampling variance within lines within generations. 
 This interaction was chosen as the appropriate error in testing the 
 main effects and the individual comparisons resulting from their break- 
 down. 
 
 The seven degrees of freedom for the means of eight generations 
 were broken down into a linear component and deviations from 
 linearity. A major portion of the variance can be attributed to the 
 nearly linear increase in the mean popping expansion over the eight 
 generations. However, a highly significant deviation exists that is not 
 accounted for by the linear trend. 
 
 The nature of the highly significant differences in performance of 
 the five lines is established by the orthogonal breakdown based on the 
 relationship of the lines (see opposite page). The significance and non- 
 significance of the comparisons exactly fit the picture of relationship 
 among the lines. 
 
 It would appear that even though the lines were grown in con- 
 tiguous plots, a certain degree of identity was maintained throughout 
 the eight-year experimental period. There must have been a certain 
 amount of flow or exchange of genetic factors conditioning popping 
 expansion between the five populations. However, the line identity does 
 indicate that some difference in gene frequency must have been main- 
 tained as a cryptic population characteristic. This would suggest that 
 breeding methods designed to recombine divergent populations derived 
 from some type of a recurring selection program might result in more 
 rapid gain and even greater increases than the method herein reported. 
 
12 
 
 BULLETIN No. 616 
 
 [September, 
 
 Table 4. Means and Standard Errors of Measurements Taken on Ears 
 Selected and Tested for Popping Expansion for 15 Generations 
 
 
 
 Measurements of tested ears 
 
 Year 
 
 Num- 
 her of 
 
 Popping 
 
 Ear 
 
 Ear 
 
 Ear 
 
 Weight of 
 
 Shelling 
 
 tested 
 
 ears 
 
 expansion 
 
 length 
 
 width 
 
 weight 
 
 shelled corn 
 
 per- 
 
 
 
 (volumes) 
 
 (inches) 
 
 (inches) 
 
 (grams) 
 
 (grams) 
 
 centage 
 
 1939 
 
 201 
 
 22.25 .342 
 
 4.99 .037 
 
 1.62 + .012 
 
 99.5 1.21 
 
 82.1 + 1.07 
 
 82.3 .21 
 
 1940 
 
 200 
 
 21.69 .337 
 
 4.56 .033 
 
 1.63 .010 
 
 99.2 1.08 
 
 80.9 0.91 
 
 81.6.23 
 
 1941 
 
 310 
 
 26.01 .203 
 
 4.39 + .027 
 
 1.57 + .015 
 
 93.9 0.76 
 
 75.5 0.64 
 
 80.5 .17 
 
 1942 
 
 310 
 
 24.58 .299 
 
 4. 15 .030 
 
 1.65 + .009 
 
 96.1+0.87 
 
 77.9 0.74 
 
 81.1.19 
 
 1943 
 
 222 
 
 26.61 + .179 
 
 4. 10 .031 
 
 1.16 + .007 
 
 80.4 + 0.77 
 
 63.8 0.63 
 
 79.4 .26 
 
 1944 
 
 310 
 
 28.06 .208 
 
 4.72 .023 
 
 1.62 + .009 
 
 89.40.71 
 
 72. 3 1.12 
 
 79.8 .18 
 
 1945 
 
 320 
 
 23.43 + .289 
 
 4.66 + .020 
 
 1.77 + .007 
 
 107.3 0.78 
 
 85.8 + 0.66 
 
 79.9 .20 
 
 1946 
 
 300 
 
 28.51 + .220 
 
 4.63 .028 
 
 1.15 + .011 
 
 118.60.79 
 
 94.2+0.60 
 
 79.4 + .20 
 
 1947 
 
 300 
 
 29.68 + .146 
 
 4.5 7 + .029 
 
 1.67 + .008 
 
 86.9 0.69 
 
 67.1 0.60 
 
 77.1 .22 
 
 1948 
 
 300 
 
 30.17 .252 
 
 5.27 + .022 
 
 1.63 .007 
 
 107. 2 0.67 
 
 86.0 0.60 
 
 80.3 .19 
 
 1949 
 
 300 
 
 34.07 .201 
 
 4.96 + .020 
 
 1.66 + .006 
 
 104.6 + 0.63 
 
 86.3 0.56 
 
 82.4.21 
 
 1950 
 
 300 
 
 35.61 + .236 
 
 5. 11 .020 
 
 1.68 +.006 
 
 109.7 0.82 
 
 90.5 + 0.65 
 
 82.8.22 
 
 1951 
 
 300 
 
 34.04 .2 10 
 
 5. 21 .023 
 
 1.65 + .005 
 
 104.2+0.60 
 
 82.3 + 0.56 
 
 78.9 .22 
 
 1952 
 
 300 
 
 37. 81 .144 
 
 4.90 + .021 
 
 1.64 +.006 
 
 91.7 + 0.67 
 
 71.3 + 0.53 
 
 77.8 + .23 
 
 1953 
 
 300 
 
 35.83 .180 
 
 4.80 + .022 
 
 1.53 + .005 
 
 75.5 0.58 
 
 56.4 + 0.49 
 
 74.7 + .31 
 
 Relationship of popping expansion to supplementary ear charac- 
 teristics. The length, width, and weight of ear; weight of shelled corn; 
 and the shelling percentage were measured on each ear tested for 
 popping expansion (Table 4). The individual values for popping ex- 
 pansion were statistically correlated with the supplementary measure- 
 ments to determine the extent of association existing between these 
 variables (Table 5). No consistent relationship exists between popping 
 expansion and ear length. Ear width, ear weight, weight of shelled 
 
 Table 5. Summary of Correlation Coefficients for Popping Expansion 
 
 and Supplementary Characteristics of Tested Ears Selected 
 
 for Popping Expansion for 15 Generations 
 
 Year 
 tested 
 
 Number 
 of ears 
 
 Supplementary measurements of tested ears 
 
 Ear 
 length 
 
 Ear 
 width 
 
 Ear 
 weight 
 
 Weight of Shelling 
 shelled corn percentage 
 
 1939 
 
 201 
 
 -.177 
 
 -.220*' 
 
 -.333* 
 
 -.388* -.383** 
 
 1940 
 
 200 
 
 + .011 
 
 -.236** 
 
 -.318* 
 
 -.424* 
 
 -.466** 
 
 1941 
 
 310 
 
 + .075 
 
 + .018 
 
 -.126* 
 
 -.232* 
 
 -.432** 
 
 1942 
 
 310 
 
 + .097 
 
 -.071 
 
 -.132* 
 
 -.246* 
 
 -.464** 
 
 1943 
 
 222 
 
 + .013 
 
 -.081 
 
 -.133 
 
 -.207* 
 
 -.166* 
 
 1944 
 
 310 
 
 + .030 
 
 -.145* 
 
 -.062 
 
 -.008 
 
 -.282** 
 
 1945 
 
 320 
 
 -.209" 
 
 -.128* 
 
 -.315* 
 
 -.384* 
 
 -.263** 
 
 1946 
 
 300 
 
 + .184* 
 
 -.204** 
 
 -.173* 
 
 -.239* 
 
 -.142* 
 
 1947 
 
 300 
 
 + .033 
 
 + .116* 
 
 + .141* 
 
 + .149* 
 
 + .084 
 
 1948 
 
 300 
 
 + .096 
 
 -.099 
 
 -.332* 
 
 -.405* 
 
 -.382** 
 
 1949 
 
 300 
 
 -.035 
 
 - .034 
 
 -.193* 
 
 -.277* 
 
 -.249** 
 
 1950 
 
 300 
 
 -.009 
 
 -.134* 
 
 -.209* 
 
 -.302* 
 
 -.191** 
 
 1951 
 
 300 
 
 + .141* 
 
 -.186** 
 
 -.289* 
 
 -.423* 
 
 -.425" 
 
 1952 
 
 300 
 
 + .197" 
 
 -.063 
 
 -.022 
 
 -.061 -.058 
 
 1953 
 
 300 
 
 + .073 
 
 -.145* 
 
 -.074 
 
 -.111 -.087 
 
 Average 
 
 4273 
 
 + .040 
 
 -.102** 
 
 -.167*" 
 
 -.240** -.263** 
 
 * Exceeds 0.05 level of significance. 
 ** Exceeds 0.01 level of significance. 
 
1957} SELECTION FOR IMPROVED POPPING EXPANSION 13 
 
 corn, and shelling percentage are all negatively correlated with popping 
 expansion. Although the relationships are quite consistent over the 15 
 generations and the over-all averages are statistically significant, the 
 correlations are not large enough to be of much value in selection or 
 prediction. 
 
 Data were also taken on shape of kernels on the individual tested 
 ears. The ears were classified as having either pointed or rounded 
 kernels. The number of ears and mean popping expansion within the 
 two classes were summarized for the years in which comparisons could 
 be made (Table 6). It is clearly evident that ears with pointed kernels 
 gave expansions consistently superior to those for ears with rounded 
 kernels. The significance of this difference is confirmed by the highly 
 significant t value for the comparison (Table 6). 
 
 Table 6. Comparison of the Mean Popping Expansion of Ears 
 Classified as to Shape of Kernel Pointed or Rounded 
 
 Pointed kernels 
 
 Rounded kernels 
 
 Difference 
 
 Voar 
 
 
 
 
 
 
 i ' < i r 
 
 tested 
 
 Number 
 
 Mean 
 
 Number 
 
 Mean 
 
 popping 
 
 
 of 
 
 expansion 
 
 of 
 
 expansion 
 
 expansion 
 
 
 ears 
 
 (volumes) 
 
 ears 
 
 (volumes) 
 
 (volumes) 
 
 1940 
 
 138 
 
 22.25 
 
 62 
 
 20.58 
 
 1.67 
 
 1941 
 
 292 
 
 26.05 
 
 18 
 
 24.76 
 
 1.29 
 
 1942 
 
 195 
 
 25.75 
 
 115 
 
 22.57 
 
 3.18 
 
 1943 
 
 171 
 
 26.83 
 
 51 
 
 25.73 
 
 1.10 
 
 1944 
 
 210 
 
 28.62 
 
 100 
 
 26.90 
 
 1.72 
 
 1945 
 
 266 
 
 23.88 
 
 54 
 
 21.20 
 
 2.68 
 
 1946 
 
 290 
 
 28.63 
 
 10 
 
 26.61 
 
 2.02 
 
 1947 
 
 295 
 
 29.69 
 
 5 
 
 29.12 
 
 0.57 
 
 1949 
 
 217 
 
 34.46 
 
 83 
 
 33.06 
 
 1.40 
 
 1951 
 
 209 
 
 34.65 
 
 91 
 
 32.63 
 
 2.02 
 
 1952 
 
 283 
 
 37.87 
 
 17 
 
 36.92 
 
 0.95 
 
 Means and 
 
 
 
 
 
 
 totals 
 
 2566 
 
 28.97 
 
 606 
 
 27.28 
 
 1.69 
 
 ,- 152 
 
 .230 
 
 = 7.35** with 10 degrees of freedom 
 
 ** Exceeds 0.01 level of significance. 
 
 Discussion and Conclusions 
 
 The aim of breeding work is the formation of types of plants 
 adapted to specific conditions. Generally speaking, little is actually 
 known about the extent of variability of plant material. Data are ac- 
 cumulating that indicate the existence of greater variability of plant 
 material than had previously been thought possible. Mather (24) di- 
 vided variability into two components: (1) free variability which mani- 
 fests itself phenotypically and (2) potential variability which is released 
 slowly only by recombination of gene complexes governing quantitative 
 characters. 
 
14 BULLETIN No. 616 [September, 
 
 The data presented here appear to be a good example of the release 
 of potential variability. A classical example of such a release is the 
 experiment conducted at the Illinois Agricultural Experiment Station 
 on selection for both high and low oil and protein content in corn. The 
 results of selection for popping expansion confirm in all essential re- 
 spects those reported by Woodworth et al. (29) in their summarization 
 of fifty generations of selection for oil and protein content. They con- 
 cluded that progress in the direction of selection apparently was still 
 being made in the High Oil and Low Protein strains, while little 
 progress had been made in either High Protein or Low Oil in the last 
 
 15 to 20 generations of selection. Two generations of reverse selection, 
 begun in 1948, have indicated that High Oil, High Protein, and Low 
 Protein may still possess considerable genetic variability. 
 
 Variability in popping expansion, as measured by both the coeffi- 
 cient of variation and the Weinberg formula, has significantly de- 
 creased over the 15 generations of selection (Fig. 2). Woodworth et al. 
 (29) reported that variability for high oil, low oil, and high protein 
 content had either increased or remained virtually unchanged over the 
 50 generations of selection. The trend of variation in response to selec- 
 tion for low protein was downward, and thus in agreement with the 
 popcorn results. 
 
 To some extent the reduction of variability over the 15 generations 
 of selection can be attributed to the gradual elimination of the types 
 giving low expansions from the effective breeding population (see 
 Table 1). In general, the elimination of undesirable genotypes of the 
 character under selection is similar to that obtained by selfing and 
 selection. The recurring type of selection program would be slower in 
 bringing this about, but complete fixation would be avoided. Since the 
 recurrent type of selection allows for recombination, new raw ma- 
 terial for positive selection is constantly being created. It is only 
 through some such breeding system that a selected population can be 
 advanced to a point where it may exceed the range of the original 
 unselected population, as has been herein demonstrated (see Table 1 
 and Fig. 1). 
 
 One may conclude from these data that the number of genes con- 
 ditioning popping expansion is relatively smaller than that for either 
 protein or oil content in corn. The inheritance of popping expansion 
 must be considerably more complex than Crumbaker et al. (7) and 
 Johnson and Eldredge (21) suggest. The level of popping expansion 
 in their studies was relatively low compared with the population 
 means of the last five generations of selection in the current study. The 
 
1957} SELECTION FOR IMPROVED POPPING EXPANSION 15 
 
 number of genes differentiating popcorn from other types of corn may 
 be relatively few in number. However, the number of genes condition- 
 ing quantitative differences in popping expansion must be of a much 
 higher order. 
 
 The consistent negative association of popping expansion and ear 
 width, ear weight, weight of shelled corn, and shelling percentage can 
 be interpreted to confirm that reported by Willier and Brunson (28) 
 and Lyerly (23). Even though these negative relationships are rela- 
 tively small, they should be considered in a popcorn breeding program 
 since ear and kernel measurements are components of yield. In this 
 instance it would appear that expansion of the popped kernels can be 
 increased by selection, but at the expense of other yield components. 
 However, it might be expected that there could be specific combina- 
 tions of inbreds that would result in hybrids not exhibiting a negative 
 relationship between popping expansion and the other components of 
 grain yield. 
 
 The consistent relationship of kernel shape and popping expansion 
 is also of interest from the standpoint of improving popcorn by plant 
 breeding. In this instance those individuals with pointed kernels con- 
 sistently outyielded the rounded types. This observed difference may 
 possibly be explained by the difference in shape and conformation of 
 the popped kernels of the two types. The pointed kernels very fre- 
 quently expand into what is commonly termed the "butterfly type," 
 which has wing-like projections from the sides of the popped kernel. 
 Within this material this type of expansion appeared to be much less 
 frequent among the rounded kernels. Rounded kernels usually form 
 the more compact, "mushroom" type of popped kernel. Differences 
 in frequency of the types of popped kernels could readily alter the 
 volumetric readings of the expanded kernels. Whether such a relation- 
 ship would exist in types of popcorn other than that used in this experi- 
 ment is not known. 
 
 Considerable attention has been given recently to new breeding and 
 selection systems that maximize the frequency of desirable genes. 
 Sprague and Brimhall (26) point out that under a system of inbreeding 
 and selection within inbred lines, a potential ceiling is established at 
 the time of the first selfing, determined by the genotype of the plants 
 selected. It seemed plausible to them that a much higher potential 
 ceiling might be established by a somewhat different approach. They 
 outlined a general approach to the problem and tentatively termed the 
 new system "truncation selection." The essential features are as fol- 
 lows: evaluate a series of individual plants for a given character, 
 
16 BULLETIN No. 616 [September, 
 
 truncate the frequency distribution at some desired level, and inter- 
 cross the individuals comprising the truncated tail. This recombination 
 would then serve as source material for a new cycle of selection. 
 
 Considerable similarity exists between truncation selection and the 
 selection method employed in the study under discussion. The selection 
 methods reported by Woodworth et al. (29) and Brunson (2) are also 
 very similar to the truncation-selection method. A breeding system 
 similar to that proposed by Sprague and Brimhall (26) had been sug- 
 gested previously by East and Jones (9) and Hayes and Garber (15). 
 Maternal-line selection proposed by Fryer (12) to improve seed pro- 
 duction of alfalfa and the fixation of selected characters in partially 
 isolated populations of side-oats grama as used by Harlan (14) also 
 bear similarity to the above-cited systems of selection. Selection of a 
 type similar to that cited above has proved effective with animals as 
 well as plants. Craig et al. (6) report that selection of a recurrent type 
 was effective for high and low weights in Hampshire swine at 154 and 
 180 days of age. 
 
 The above-mentioned systems all embody the basic principles of 
 recurrent selection. Hull ( 18) in 1945 proposed a procedure which he 
 called recurrent selection for specific combining ability. Jenkins (19), 
 however, published a detailed description of the recurrent-selection 
 method in 1940. The two differ in that Hull's method used a homozy- 
 gous tester instead of a heterozygous tester proposed by Jenkins. Corn- 
 stock et al. (5) suggested a system whereby recurrent selection could 
 be carried out reciprocally. 
 
 Much effort has been expended in the last decade on testing the 
 effectiveness of systems of recurrent selection. Hayes et al. (16) and 
 Johnson (20) present a good review of literature and discuss the utili- 
 zation of recurrent selection in plant breeding. Johnson (20) classified 
 the systems of recurrent selection into three rather distinct categories, 
 based on objectives and the resulting changes in gene frequency for 
 any given character: recurrent selection for specific combining ability, 
 recurrent selection for general combining ability, and recurrent selec- 
 tion for specific phenotypic characters. 
 
 The similarity between recurrent selection for a specific phenotypic 
 character and controlled mass selection was pointed out by Johnson 
 and Goforth (22). They stated that for characters with a relatively 
 high heritability, a system of mass selection in which only desired 
 phenotypes are permitted to interpollinate might prove to be an effec- 
 tive yet simple method of plant breeding. That such a breeding pro- 
 gram would be successful is substantiated by the currently reported 
 
1957] SELECTION FOR IMPROVED POPPING EXPANSION 17 
 
 results, since the increased level of popping expansion resulted from a 
 selection method basically similar to controlled mass selection. It is of 
 interest that Clary (4) suggested that a program of mass selection, 
 followed by recurrent selection either in the original open-pollinated 
 varieties of popcorn or in a synthetic variety made from tested inbred 
 lines, should result in isolation of lines superior in popping expansion 
 to those he investigated. 
 
 The similarity between controlled mass selection, truncation selec- 
 tion, and recurrent selection for a specific phenotypic character is 
 rather obvious. Actually it is difficult to make a concise differentiation 
 between these methods. They may differ slightly in operational pro- 
 cedures with the various crops for which they may be employed and in 
 the extent of the selection pressure exerted in each instance on the 
 character under selection. The three methods should yield essentially 
 the same end results. 
 
 The method herein reported cannot be considered terminal in breed- 
 ing a crop such as popcorn in which hybrids are the most acceptable 
 form for commercial production. It should be considered a means of 
 maximizing the frequency of desirable genetic factors and integrated 
 subsequently with the regular methods of inbred production and im- 
 provement. If the genetic factors involved are largely additive in effect, 
 inbred lines derived from this material should possess a potentially 
 higher level of effective genetic factors than those isolated from the 
 original population. 
 
 Summary 
 
 Selection for improved popping expansion through IS generations 
 was initiated in 1939 and completed in 1953. Individual ears were tested 
 for popping expansion and only those with the highest expansions were 
 planted to produce the next generation. The method of selection was a 
 recurrent type similar to those termed modified mass selection or 
 truncation selection. 
 
 The selection method employed proved effective in increasing the 
 level of popping expansion within the population. The mean popping 
 expansion was 22.2 volumes for the initial population and increased to 
 35.8 volumes at the end of the selection period. The mean increase in 
 popping expansion over the last eight generations was shown to be very 
 nearly linear. The magnitude of the lowest and highest variates also 
 showed a corresponding increase throughout the selection period. The 
 lowest variate increased from 5.6 in 1939 to 26.0 volumes in 1953, while 
 the highest variate increased from 31.6 to 44.0 volumes for the same 
 
18 BULLETIN No. 616 [September, 
 
 period. Cryptic differences in popping expansion appeared to be main- 
 tained among the five lines established in 1946 and continued to the end 
 of the experimental period, even though opportunity for genetic recom- 
 bination existed between the five lines. Variability in popping expansion 
 as measured by the coefficient of variation and the Weinberg formula 
 has decreased significantly over the 15 generations as a result of 
 selection. 
 
 A consistent negative correlation between popping expansion and 
 the width and weight of ear, weight of shelled corn, and the shelling 
 percentage was demonstrated, but the association was not considered 
 sufficiently large to be of significant value in selection or prediction. No 
 consistent relationship exists between popping expansion and ear 
 length. 
 
 Individual selected ears were classified as having either pointed or 
 rounded kernels. Those ears with pointed kernels on the average 
 yielded 1.69 volumes higher expansion than those with rounded kernels. 
 
 This selection method is not considered terminal, but rather a means 
 of maximizing the frequency of desirable genetic factors. It is proposed 
 that such a system would be of value for increasing the frequency of 
 genes that are largely additive in effect before employing the usual 
 methods of inbred production and improvement. Fixation by inbreed- 
 ing should establish a potentially higher level of effective genetic factors 
 from the selected population than would prove possible from the 
 original population. 
 
 Literature Cited 
 
 1. BARTLETT, M. S. Some examples of statistical methods of research in agricul- 
 
 ture and applied biology. Jour. Roy. Stat. Soc. Suppl. 4:137-183. 1937. 
 
 2. BRUNSOX, A. M. Popcorn: selecting for added popping expansion would pay 
 
 large growers. U. S. Dept. Agr. Yearbook of Agr. 1931:441-443. 1931. 
 
 3. BRUNSON, A. M., and SMITH, G. M. Popcorn. U. S. Dept. Agr. Farmers' Bui. 
 
 1679. 18p. 1948. 
 
 4. CLARY, G. A study of the inheritance of expansion in popcorn. Ph.D. thesis. 
 
 Purdue Univ. 81p. 1954. 
 
 5. COMSTOCK, R. E., ROBINSON, H. F., and HARVEY, P. H. A breeding procedure 
 
 designed to make maximum use of both general and specific combining abil- 
 ity. Agron. Jour. 41:360-367. 1949. 
 
 6. CRAIG, J. V., NORTON, H. W., and TERRILL, S. W. A genetic study of weight 
 
 at five ages in Hampshire swine. Jour. Anim. Sci. 15:242-256. 1956. 
 
 7. CRUMBAKER, D. E., JOHNSON, I. J., and ELDREDGE, J. C. Inheritance of popping 
 
 volume and associated characters in crosses between popcorn and dent corn. 
 Agron. Jour. 41:207-212. 1949. 
 
 8. DEXTER, S. T. Conditioning popcorn to the proper moisture content for best 
 
 popping. Mich. Agr. Exp. Sta. Quart. Bui. 29:64-69. 1946. 
 
1957] SELECTION FOR IMPROVED POPPING EXPANSION 19 
 
 9. EAST, E. M., and JONES, D. F. Genetic studies on the protein content of maize. 
 Genetics 5:543-610. 1920. 
 
 10. ELDREDGE, J. C. The story of better popcorn. Popcorn Merchandiser 6(10) :4,6, 
 
 37. 1952. 
 
 11. ELDREDGE, J. C., and LYERLY, P. J. Popcorn in Iowa. Iowa Agr. Exp. Sta. 
 
 Bui. P54: 753-778. 1943. 
 
 12. FRYER, J. R. The maternal-line selection method of breeding for increased 
 
 seed-setting in alfalfa. Sci. Agr. 20:131-139. 1939. 
 
 13. GRISSOM, D. B. Heritability and association of characters affecting popping 
 
 volume in dent-corn popcorn crosses. M.S. thesis. Iowa State Col. 1951. 
 
 14. HARLAN, J. R. The breeding behavior of side-oats grama in partially isolated 
 
 populations. Agron. Jour. 42:20-24. 1950. 
 
 15. HAYES, H. K., and CAREER, R. J. Synthetic production of high protein corn in 
 
 relation to breeding. Amer. Soc. Agron. Jour. 11:309-318. 1919. 
 
 16. HAYES, H. K., IMMER, F. R., and SMITH, D. C. Methods of plant breeding. 
 
 2d ed. McGraw-Hill Book Co. 551p. 1955. 
 
 17. HUELSEN, W. A., and BEMIS, W. P. Artificial drying and rehydration of pop- 
 
 corn and their effects on popping expansion. 111. Agr. Exp. Sta. Bui. 593. 
 68p. 1955. 
 
 18. HULL, F. H. Recurrent selection for specific combining ability in corn. Amer. 
 
 Soc. Agron. Jour. 37:134-145. 1945. 
 
 19. JENKINS, M. T. The segregation of genes affecting yield of grain in maize. 
 
 Amer. Soc. Agron. Jour. 32:55-63. 1940. 
 
 20. JOHNSON, I. J. Effectiveness of recurrent selection for general combining 
 
 ability in sweetclover, Melilotus officinalis. Agron. Jour. 44:476-481. 1952. 
 
 21. JOHNSON, I. J., and ELDREDGE, J. C. Performance of recovered popcorn inbred 
 
 lines derived from outcrosses to dent corn. Agron. Jour. 45:105-110. 1953. 
 
 22. JOHNSON, I. J., and GOFORTH, F. Comparison of controlled mass selection and 
 
 recurrent selection in sweetclover, Melilotus officinalis. Agron. Jour. 45:535- 
 539. 1953. 
 
 23. LYERLY, P. J. Some genetic and morphologic characters affecting the popping 
 
 expansion of popcorn. Amer. Soc. Agron. Jour. 34:986-999. 1942. 
 
 24. MATHER, K. Polygenic inheritance and natural selection. Biol. Rev. 18:32-64. 
 
 1943. 
 
 25. SMITH, G. M., and BRUNSON, A. M. Hybrid popcorn in Indiana. Ind. Agr. 
 
 Exp. Sta. Bui. 510:18p. 1946. 
 
 26. SPRAGUE, G. F., and BRIMHALL, B. Relative effectiveness of two systems of 
 
 selection for oil content of the corn kernel. Agron. Jour. 42:83-88. 1950. 
 
 27. WEINBERG, W. Zur Messung der Streuung bei extensiven Masszahlen. Arch. 
 
 Rassen. u. Gesell. Biol. 18:84. 1926. 
 
 28. WILLIER, J. G., and BRUNSON, A. M. Factors affecting the popping quality of 
 
 popcorn. Jour. Agr. Res. 35:615-624. 1927. 
 
 29. WOODWORTH, C. M., LENG, E. R., and JUGENHEIMER, R. W. Fifty generations 
 
 of selection for protein and oil in corn. Agron. Jour. 44:60-65. 1952. 
 
UNIVERSITY OF ILLINOIS-URBANA