A Biometrical Study of the Influence of Size of Brood Cell Upon the Size and Variability of the Honeybee (Apis mellifera L.) by Roy A. Grout, 1931

Naturalists and observers have been interested in the size, shape and types of brood cells of a bee colony for many centuries. These studies date back at least to the fourth century B.C. to that ancient pioneer in the development of science, Aristotle. In a review of the literature of this subject we must also mention such outstanding naturalists as Swammerdam, Reamur, Maraldi and the mathematician, Koenig, who observed the marvelously consistent shape of the cells of the honey comb and studied the mathematics of its structure.

According to Ruber (31), Schirach, who discovered that the worker bee was possessed of the same sex as the queen bee, stated that nothing but certain physical conditions, such as a special food and a more spacious lodging, was needed for the worker bee larvae to become real queens. Desiring the support of an eminent philosopher in his contentions, he corresponded with Bonnet, who in turn corresponded with Huber, who confirmed Schirach's results.

It was the great naturalist, Huber (32), who, in 1791, at the suggestion of Bonnet, first attempted to investigate the effect of the size of cells upon the size of the honeybee. He removed from a hive all combs containing worker cells, leaving only drone combs. Moths invaded the hive, however, and ruined his first experiment. On his second attempt, he observed that the queen finally oviposted with reluctance in the drone comb, but that on the forth day the bees disappeared from their cells. Upon placing in the hive a frame containing sealed drone brood in small cells, he observed that the bees set about to remove the drone brood from the cells so that the queen might have a suitable place to ovipost. Ruber next attempted to rear worker bees in drone cells by grafting the worker bee larvae in the cells of a drone comb by removing the drone larvae and placing them with 1 day old worker larvae. This experiment was successful, but Ruber, upon examining the pupa in an advanced stage, did not recognize any difference in the size of the bees. He then repeated the experiment using drone larvae instead of worker larvae. This time he observed that the drones reared in the drone cells were larger than those reared in worker cells.

Many observers since Huber have conducted similar experiments and have observed the phenomenon of worker brood reared in drone cells sealed with level cappings. Thus, Alley (1), in 1869, records placing a queen and three pounds of bees on drone combs with the result that worker bees were reared in level-capped drone cells. He does not state, however, whether he noticed any changes in the size of the emerging brood. Gundelach, according to Michailov (43), placed a swarm of bees in a glass-walled observation hive on drone comb and observed worker bees emerging from the large cells. Berlepsch (9), in 1876, conducted a similar experiment and confirmed the results of Gundelach. He also cited a similar result obtained by Bessel. Other investigations cited by Michailov (43) were made by Gunther, Klempin, Zarudski, Zesselski, Lehzen and Hanneman, and Buttel-Reepen. Pincot, according to Gillet-Croix (26), investigated the phenomenon at the biginning of the present century and Getaz (25) cites Drory of France and a similar case in Germany. None of the above observers, however, investigated the size of the bee by microscopical examination but depended entirely upon visual examination. Only Zarudski claimed that there was any increase in the size of worker bees due to their rearing in drone cells.

The first microscopical examination of the chitinous parts of worker bees reared in drone cells was conducted in 1901 by Martynov (41), who measured 100 proboscides of bees from the apiary of the Moscow Institute. He determined that the average length of proboscis of worker bees emerging from normal worker cells was 6.06 mm., while the length of proboscis of those emerging from drone cells was 7.01 mm., showing an increase of 0.95 mm.

A similar and more extensive investigation was undertaken by Michailov (43), in 1925, who chanced to find in one of his colonies a drone comb containing worker brood sealed with level cappings. Taking advantage of this opportunity, Michailov, who had previously stated that there was no great difference in the size of worker bees reared in drone cells, made a microscopical examination of 6 characteristics of 200 bees, 100 taken from the drone comb and an equal number from a normal brood comb, taken at approximately the same time. He determined by statistical methods that the bees reared in drone cells were significantly larger than their worker-cell sisters and drew the following conclusions in this respect:
  1. That worker bees reared in drone cells weigh 11.36% more than worker bees reared in worker cells.
  2. That the proboscis increases 4.83% in length due to the effect of the larger cells.
  3. That the increase in the size of cell gives a corresponding increase in the size of the right fore wing, an increase in length of 2.69% and an increase in width of 2.06%.
  4. That the increase of the size of the cell causes an increase in the sum of the widths of the third and fourth tergites of 4.37%.
  5. There is no significant differences in the average number of hooks on the right hind wing between the two groups of bees.
  6. That the bees reared in the drone cells are decidedly more variable than their worker-cell sisters except in the case of the number of hooks on the right hind wing where the variation is consistant in both groups and therefore unrelated to size of cell.

A further pursuit of the literature relating to the effect of the size of the brood cell upon the size and variability of the honeybee leads into several more or less distinct controversies and studies which bear upon the subject from various angles. The first of these controversies concerns itself with an extensive discussion of the length of proboscis and its relation to honey storing ability with special reference to ability to forage upon and pollinate red clover (Trifolium pratense). The second controversy has to do with the effect of the age of the brood comb upon the size of the honeybee. Both of the above controversies have influenced and have operated to bring about a controversy concerning the use of artificial foundation with an enlarged cell base. The last controversy is concerned largely with the study of the variability of social insects due to the variability of the brood cells. Since the above four controversies occurred at approximately the same time, the writer feels that the best method of approach to a review of their literature is to discuss each subject separately.

1. Length of proboscis and its relation to honey storing ability.

The length of proboscis, its relation to honey storing ability and, in particular, its relation to the acquisition of nectar from red clover rose to a peak in this country at the very beginning of the present century. Previous to that time a few observers had noticed honeybees working on red clover, and Rankin (56) had successfully attempted to breed a strain of bees having long tongues. These remarkable traits soon disappeared, however. Root (59) was the first great disciple of the long-tongued worker bee. He discovered in one of his apiaries a colony which was working on red clover. Upon measuring the tongue reach of these bees, he discovered that this colony had an unusually long tongue reach, 0.21 inches, whereas the average tongue reach was only 0.16 inches. He continued by raising queens from the mother of this colony and sold them for "red clover queens". This desirable quality shortly disappeared due to the inability to control the queen's mating. It is of interest to note that Kulagin (30) measured the length of proboscis of ten bees that were the progeny of four Root queens that had been sent to Russia by Titoff and found that the average length was 6.22 mm. as compared to an average of 6.21 mm. for the common black bees of Central Russia.

The contentions of Root regarding the tongue reach of these bees and their ability to acquire nectar from red clover raised a voluminous controversy which lasted until the middle of the year 1902 when it was dropped as suddenly as it commenced. Beekeeping savants such as Miller (46), Doolittle (22), Dadant (16), Gillette (27), Getaz (25), Cook (15), Swarthmore (64) and many others investigated and wrote concerning the subject. The greater part of the above mentioned believed that the longer proboscis was directly related to a greater storing ability in spite of the fact that very little scientific research was undertaken. The importance of their controversy lies in their contributions to the technique of measurement, the breeding and selection of bees and the influence of their investigations upon further scientific study along similar lines.

Previous to this time, Wankler (70) of Germany, had attempted to breed for length of proboscis and had invented and used by 1882 an instrument for determining the length of the bee's tongue. According to Gotze (29), Wankler was the first to show that the bees of different races may differ in the length of their respective proboscides. Likewise, Charton (13), of France, had invented the Charton glossometer in 1892 and by 1897 had presented some figures which seemed to indicate that bees store in proportion to the length of their proboscides.

In Russia, scientific workers have recognized the importance of the relation of the honeybee to the pollination of the red clover and have, for many years, made a very extensive study of the length of the proboscis of the honeybee and the relation of honeybees to seed production. The Russian territory is well suited to such a study since hybridization and shipping of bees from one part of the territory to another has practically never occurred. Klingain, according to Michailov (45), in experiments conducted from 1908 to 1913, showed in his first experiment that bumblebees pollinated 49.4% of the flowers in comparison to 45% in the case of Caucasian bees. In his second experiment he found that bumblebees pollinated 46% while the honeybees pollinated 31.1% of the flowers. He further calculated that it would be necessary to have one colony of Caucasian bees per acre of red clover in order to insure proper pollination.

Chochlov (14) stated that the minimum average length of the red clover corolla tube was 8.34 mm., that the bees push a part of their heads a distance of 0.65 mm. into the corolla tube, and that the nectar rises in the corolla tube a distance of 1 mm. Thus a bee would have to have a proboscis reach of 6.69 mm. to acquire nectar from red clover. Chochlov also found that the Abkhasian bee and the Kars bee of the Caucasus had a length of proboscis equal to 6.69 mm. The essential part of the work done by Chochlov is his establishing for the first time a technique for preparing the proboscis for measurement. His method consisted of anesthetizing with chloroform, killing in boiling water, boiling in potassium hydroxide solution (KOH) and preserving in oil of cloves after the parts had been washed in water for several days. Ewert (23) conducted a similar investigation of the length of the proboscis in relation to the depth of the corolla tube of red clover and essentially confirmed the results obtained by Chochlov.

A more extensive and accurate work on red clover was conducted by Gubin (30), who, contrary to Chochlov and Ewert, showed that the length of the proboscis would have to be from 7.9 mm. to 8.9 mm. in order to reach the nectar in the corolla tube. Since the bees with the greatest length of proboscis were found by Skorikow (63) to be Caucasian bees having a proboscis length of 7.55 mm., Gubin concludes that real red clover bees do not exist in Russia. He also showed that, in using the technique for preparing the proboscis for measurement as set forth by Choclov, the proboscis and other chitinized parts of the skeleton shrink, particularly upon boiling in potassium hydroxide solution, and that the submentum shrinks most (6.31%), the ligula, 3.82% and the mentum only 1.50%. He further showed that by placing the material in a 30% alcohol solution and running it up to a 70% alcohol solution, the shrinkage seldom surpasses 1.5%. Gorbatscheff (28), in 1929, differed with Gubin and claimed the the Caucasian race is the only one which makes use of the nectar of the red clover, but, according to Gotze (29), the material presented by Gorbatscheff is not capable of complete proof.

Additional tests in Russia regarding the relation of the honeybee to the pollination of red clover were made by Manokhin and Koorochkin, according to Michailov (45). Manokhin reported an increase of tenfold in the seed production of red clover influenced by the flight of Caucasian bees in comparison with an isolated field. One plot isolated from insects produced only 180 grams of seed per hectare and another produced but 225 grams; while a similar area, located near Caucasian bees, produced 46.4 kg. and another located near black bees produced 32.5 kg. of seed. The test by Koorochkin in 1926 showed that common local bees of Northern Russia produced seed in 0.99% of the flowers, Caucasian bees produced seed in 10.42% and bumblebees produced seed in 27.93%. Free blooming clover showed seed production in 38.93% of the flowers while a plot isolated from insects showed only 0.27% pollinated.

In 1929, a Russian worker by the name of Savelyeff (62) made an extensive investigation of the effect of various methods of killing, treating and preserving the chitinized parts of the bee skeleton with special reference to the proboscis. In a comparison of two methods, namely boiling fixation plus preservation in 70% alcohol and direct preservation in alcohol without fixation, he arrives at the conclusion the the differences obtained in a composite measurement of the proboscis (submentum plus mentum plus ligula) are not great enough to be due to any reason other than error in measuring. While the mentum showed an increase over the boiling-fixation method plus preservation in 70% alcohol, the submentum showed a decrease, while the ligula showed no significant dufference in length. Thus, contrary to Gubin, Savelyeff showed that the combined measurement of the three parts showed no significant difference due to the treatment. He agreed with Gubin that the use of a coefficient or constant to correct the length of the proboscis is improbable due to the different behavior of the parts. He differs with Gubin, however, in his conclusion that the correlation among the parts of the proboscis is the same as one would find in an examination of untreated material.

Alpatov (2), in 1930, reported the result of an investigation of proboscides preserved and measured in alcohol in comparison with proboscides boiled in a 5% solution of potassium hydroxide. He found that the treatment with potassium hydroxide solution shortened the total length of the proboscis by 2.6%. At the same time he investigated the tongue length of the Mingrel bees (Apis mellifera var. caucasica Gorb.) and showed that this race had the longest proboscis of any of the races of honeybees.

In Germany, Gotze (29), in 1927, declared that the probability of error with the Russian technique was so great that it was impossible to obtain correct results. In consequence, he measured the second member of the labial palpi as an estimation of the length of the proboscis. He also made a scientific examination of the ability of the bees to acquire nectar from several varieties of clover. Following the precedent of Ewert, he did not measure the ovary since the nectar stands above it if nectar is present in sufficient quantity. He found that if the nectar rising in the corolla tube reached a depth of 7.25 to 7.5 mm., the bees of one of his colonies, number 47, could regularly acquire nectar, while the rest of his bees could not work red clover unless the nectar rose beyond this point or the bee had a proboscis longer than the mean of its colony. He concluded that, in general, even those stocks having the longest average length of proboscis do not meet the requirements of clover storing ability; that, with each increase in the length of the proboscis, the ability to acquire nectar from red clover increases markedly; and that certain existing varieties of red clover can theoretically be used by those bees having the longest known proboscides. He maintains that a modification of the length of proboscis due to climate factors has yet to be proven by experimental data, in spite of the fact that Russian workers have shown that the length of proboscis seems to decrease going from south to north.

In regard to the question of how much better use the longer proboscis will prove to be, we must mention Merrill (42), who, in 1922, investigated the relation of length of proboscis, carrying capacity and colony strength to honey storing ability. He determined that a correlation between length of proboscis alone and storing ability could not be found, but that the length of the proboscis plus carrying capacity and colony strength was highly correlated with yield. He concluded that: (1) there is a distinct correlation between length of proboscis, carrying capacity and the amount of honey stored; (2) there is a distinct relation between the number of bees found in the colony in the spring and the size of the above named physical characters; (3) that while it is very strongly indicated that it would be advantageous to a bee to excel in all three of these physical characters, yet, if she is deficient in one character, the disadvantage may be overcome if she possesses one of the other characters to a marked degree.

Hutson (33), in 1926, continued this study, working with small numbers of bees, and confirmed the above results showing that there was no marked agreement between length of proboscis and honey stored, but that there was a marked agreement between the number of bees in a colony and the yield.

Higher yields from long-tongued races have been reported rather frequently. In this connection it is well to note that Merrill and Hutson investigated only modifications of one and the same race. Zander (73), in 1919, reported unusually high yields from Caucasian bees in Germany. The constitution of the honey was at the same time totally different, and Zander believed that this race prepared the honey in a different manner since the source must have been the same. Similarly, Tiadmann (67), in 1925, tells how Cirsium oleraceum had been used by the Krain bees while the native Hannover bees could not acquire nectar from this source. According to Gotze (29), other high yields have been reported by conscientious observers such as Tuschoff and Braun in 1927, but whether these yields were based upon length of proboscis or other properties is not known.

Furthert contributions to the question concerning the acquisition of nectar from red clover have been presented at intervals over a period of years in this country by outstanding figures such as Folsom (24), Phillips (51), Dadant (17), Demuth (20), Dietz (21), Pammell (48), Robertson (58), Pellett (50), and Burrill (11). Robertson believed that the bees perforated the corolla in order to reach the nectar. Dadant confirmed this statement. Pellett believed that bees could reach the nectar in some of the flower tubes, especially in years of drouth. Pammell observed honeybees on red clover but he did not believe, after measuring the corolla tube, that the tube is shortened enough by drouth to enable the bee to reach the nectar in it. Burrill cited many ways in which the honeybee could acquire nectar from red clover by means of perforations of the corolla tube due to other insects, honeydew secreted by aphids on clover, and sap leaks.

2. Age of comb controversy.

The controversy concerning the age of comb and its possible effect upon the size of the emerging brood dates back at least to the middle of the 19th century. Quinby (54), in his book published in 1865, states that old combs are good for a period of 10 to 15 years due to the fact that while the bottoms of the brood cells are filled with cast-off pupa skins, the cocoon, excrement and varnishing resulting from each generation, the side walls are only slightly thickened and the bees lengthen them to compensate for the thickening at the midrib. It is Quinby's opinion that the cells are unquestionably built larger than necessary. Dadant (18), Miller (47), Root (60), and others took part in this controversy and all believed that the age of the comb did not materially affect the size of the emerging bees.

This controversy first originated in Europe and then spread to our country. Riedenbach (57), of Germany, showed by filling new and old combs with water that although the midrib is noticeably thickened by the emerging generations, the cell volume is not greatly changed due to the lengthening of the side walls. Ludwig (40) coincided with Riedenbach in his contentions and by actual measurement showed that there is no difference existing in the general roominess of brood cells of old and new combs. Brunnich (10) believed that the danger of the cells of old brood combs becoming smaller with age was not as great as many beekeepers believed. Rambaldi (55) reports that in 1927 he had kept Palestine and North African bees on combs built from Root foundation having 856 cell bases per square decimeter during hundreds of generations and had noticed no effect upon the size of the bees.

In Russia the study of the effect of old combs upon the size of the emerging bees was first attempted by Tuenin (68), who showed, by weighing the emerging bees from combs from which 2, 6, 28, and 38 generations had emerged previous to the experiment, that the weight decreased with the number of generations from 0.12612 gms. to 0.10695 gms. and that the cell diameter decreased from 5.262 mm. to 4.99 mm. He concluded that as the number of generations that emerged from the cells became more numerous, the resulting bees became smaller as indicated by the weight of the emerging bees.

Michailov (44) continued this study by measuring 5 physical characters of the skeleton, namely, length of proboscis, length of the right fore wing, width of the same wing, the summation of the widths of the 3rd and 4th tergites, and the number of hooks on the right hind wing. He showed that the size of the cells as reduced by the emerging generations (5.89% in the diameter due to 16 and 18 generations) is accompanied by a significant reduction in the size of bees. By reducing the cell diameter by approximately 3% there was no significant reduction in the body size of the emerging bees. the depth of the cell showed no influence upon the size of the bee. He concluded that, in order to enlarge the bee by using artificial foundation, we should pay particular attention to the diameter of the cell and not to the depth of the cell. According to Rupp (51), who based his conclusions on the work of Michailov, a comb is too old to use for brood rearing when it is three years old based upon the figure of 5 to 6 generations per year.

3. The enlarged cell controversy.

With the invention of artificial comb foundation by Mehring, in 1857, a control of the size of the cells built by honeybees was first accomplished since it was discovered that the bees would build cells with the same dimensions as the imprint of the cell base upon the artificial foundation. The importance of his invention was the elimination of excess drone comb resulting in colonies composed almost entirely of worker comb. It also initiated a study of the exact size of the cells built by bees.

According to Dadant (19), Collin measured the dimensions of cells and stated in 1865 that there were 854 cells per square decimeter. Langstroth repeated the experiment and calculated that there were 838 cells per square decimeter and Charles Dadant confirmed his results. According to Baudoux (8), the following are results concerning the size of the cells of natural comb:

The house of Fratelli Piana in Italy calculated from measurements that there were 860 cells per square decimeter; another house in Italy measured comb from three different colonies and found that there was a considerable variation in the size of cells, namely 813, 807 and 854 cells per square decimeter. Baudoux measured combs taken from two different colonies and found that while the cells of one colony measured 854 cells per square decimeter, the cells from the other measured but 807 cells per square decimeter.

Concerning the size of cells built by different races of bees, Pincot, according to Gillet-Croix (26), reports that the Italian race builds 764 cells per square decimeter, that the bees of Burgundy build 798, that the common black bee native to France builds 854 and that a "degenerated common bee" builds 924 cells per square decimeter. Halleux, in 1890, according to Szezawinski (65), calculated that the black native bees build 845 cells per square decimeter. Rambaldi (55) records the North African bee as building 940 cells per square decimeter.

Baudoux (7), of Belgium, was the first to advocate the use of artificial foundation with an enlarged cell base. In 1893, he reports that a Mr. Fromont measured natural combs and found that the greater part had 825 cells per square decimeter in comparison with certain sheets of artificial foundation which had as high as 907 cells per square decimeter. Baudoux, struck by the reduction in the size of bees from an old skep containing combs having 912 cells per square decimeter, conceived the idea of raising bees in enlarged cells. He accomplished this by means of stretching normal foundation to the size he desired and had by 1896 sufficiently proved his point in Belgium, that a manufacturing company began to place upon the market artificial foundation having an enlarged cell base. It was Baudoux's contention that the nurse bees, following a natural instinct, filled the bottom of the larger cell more copiously with larval food, that this resulted in a larger bee, He also intimated that the larger bee would generate more body heat which would result in a greater quantity of brood.

By means of stretching foundation, he experimented with various sizes of foundation having 750 cells per square decimeter, 740, 730, 710, 700 and down to 675 cells per square decimeter. By means of a glossometer he determined the tongue reach of his colonies and by means of a thoraxometer, the diameter of the thorax. He found that with an increase of 50 cells per square decimeter in the size of the foundation, there was a corresponding decrease of 0.5 mm. in the tongue reach. His thoraxometer gives a diameter of the thorax as 3.7 mm., 3.9 mm., 4.1 mm., and 4.3 mm. for the bees reared in cells built from foundation having respectively 850, 800, 750 and 700 cells per square decimeter. He arrived at the conclusion that foundation having 700 cells per square decimeter gave a bee which was superior in all its measurements to those reared in combs built from the smaller sizes.

Independent of the work done by Baudoux, Pincot, according to Gillet-Croix (26), arrived at the idea of rearing bees in enlarged cells from a slightly different angle. Noticing the difference in size of the bees from a swarm placed on foundation and the bees of the parent stock reared in natural comb, Pincot came to the conclusion that this phenomenon was due to the natural cells being larger than those drawn from the foundation and actual measurements confirmed his theory. He then started experimenting with foundation having 736 cells per square decimeter and reports that during a two-year period 30 colonies using this size of foundation gathered approximately one-third more honey than did 30 colonies on normal foundation. In 1910 his apiaries were destroyed by a flood and Pincot was forced to abandon his experiments.

Lovchinovskaya (39), of Russia, reports, after present investigations in 1930, that an investigation concerning the effect of enlarged cells upon the size and activity of the honeybee was undertaken in that country in 1925. For this purpose artificial foundation was made with an enlarged cell base of 5.85 mm. against 5.45 mm. in normal cells. The results of one experiment proceeding for 2 years with 10 colonies showed that the honeybees placed on the enlarged cells lived a normal life, both the worker bees and the queen worked normally. When only one frame of the enlarged cells was placed with nine frames of the normal size in a colony, the queen's attitude toward the enlarged cells was changed and she did not oviposit in the comb in spite of the fact that the bees worked upon it as they did on the other nine combs. When the reverse experiment was undertaken, with nine enlarged combs and one normal comb, the queen laid in the enlarged cells at once. Lovchinovskaya concludes that the worker bees regard with indifference the enlarged cells while the queen prefers the smaller cells.

He continued extensive investigations and showed that: (1) Bees from the enlarged cells weigh more than bees reared in normal cells. (2) The average weight of bees for a singlt year varies according to the conditions operating during the year. (3) The enlarged bees as well as the normal bees are heaviest during May and decrease in weight during the next three months. (4) The weight of bees leaving the hive is greatest during the first half of the day. (5) The weight of emerging bees is between 5% and 6% greater for bees emerging from the large cells. (6) That the emerging bees of both groups weigh more than the old bees. (7) In the case of bees returning to the hive, the weight of enlarged bees exceeds the weight of normal bees by 10.7%, while in the case of bees leaving the hive, the enlarged bees exceed by only 4.8%. (8) The load carried in the honey stomach of the enlarged bee is 52.6% heavier than that carried by the normal bee. (9) That the normal bee carried a load equal to 14% of her own weight while an enlarged bee carries a load equal to 20.4% of her own weight. (10) From the results of one season, the bees reared in the large cells gather more honey than those reared in normal cells, but the production cannot be judged from the data of one season.

4. Studies on the variability of the honeybee with special reference to size of brood cell.

The first paper giving data upon the variability in bees appears to be that of Koshevnikov (36), who, in 1900, studied the number of hooks on the hind wing. In 1905, the same author presented data which could be arranged in the form of a correlation table. Landacre (38), in 1901, counted the number of hooks on the hind wing but, according to Phillips (52), his data was not presented in statistical form. In 1903, Casteel and Phillips (12) made a biometrical study of the wing venation of the drone and worker bees and arrived at the conclusion that the uncreased variability of the drone was due in part to the increased variability of the cells in which they were reared.

Bachmetjew (6), in 1903, began a study of the variability of the hooks on the hind wings of honeybees. His results and conclusions have been strongly criticised in the literature, and Pearl (49), in 1910, wondered whether Bachmetjew was really serious or whether he was attempting to perpetrate a great biometrical joke. Phillips (53) recalculated the data presented by Bachmetjew in 1909 and discovered that the results were entirely normal and agreed with the results of other investigators.

Kellogg and Bell (35), in 1904, showed that there was greater variability in single wing veins than in the length or breadth of the entire wing and that there was a greater variability in the number of hooks on the hind wing than in the wing venation. This variability was as great in workers as it was in drones. Kellogg (34) later made a further investigation and concluded that, except for the number of hooks on the hind wing, drones were more variable than workers. He also found that the variability of drones reared in worker cells was greater than drones reared in drone cells, and stated that this greater variation was not due to special extrinsic factors such as size of cells.

In Russia many studies have been made upon the variability of the honeybee and the factors which influence variation. These investigations have been carried out on a large scale and have been calculated by statistical methods; they constitute a large part of our knowledge concerning the variability of the honeybee. Among these contributions are the works of Michailov, Alpatov, Tuenin, Choclov and others as cited by Alpator (3) in 1929.

Mention should also be made of biometrical studies on social insects, such as the work of Wright, Lee and Pearson (72), in 1907, on Vespa vulgaris from a single nest and by Thomson, Bell and Pearson (66), in 1909, on a general wasp population, In the first named paper, the authors re-examined the data of Casteel and Phillips and agreed that those data show a greater variability in drones than in worker bees, a condition which is reversed in the case of Vespa vulgaris. Other biometrical studies of note concerning social insects include the studies of Warren (71) on termites and the seasonal variation occurring in their forms, Alpatov and Palenitschko (4) who worked on different species of ants, and Arnoldi (5) who presented data concerning the variability of the ant Cardiocondyla stambulowi Forel.