1. 1872 normal mice bred during November 1922-October 1925 had an average litter-size of 6·18, and a male percentage of 51·7 ± ·77, both of which figures are lower than the corresponding ones found for 1921–2.

  2. The annual variation in the sex-ratio is, however, not very appreciable, the figures for the four years running 54·2 ± 1·04, 50·4 ± 3·22,52·2 ± I-18, 51·4 ± 1·09. The average size of litter during these four periods ranged from 6·65 to 5·93.

  3. Considerable seasonal variation in both fertility and sex-ratio occurred, the highest figures being fertility 6·46 and male percentage 55·9 ± 1·83 in the October-December quarter, and the lowest being fertility 5·82 and male percentage 48·2 ± 1·46 in the April-June period.

  4. Litter-size appeared to be uncorrelated with any sort of definite variation in the sex-ratio.

  5. Recent work on the mouse is discussed.

IN two papers published in 1924(11, 12) the results of normal mouse breeding for the year 1922, together with a summary of the results of other workers on this and other animals, were dealt with.

The mouse colony, which was originally founded in the early part of 1922, has been kept up during the last two years with the aid of a grant from the Government Grants Committee of the Royal Society, and during the last year has been greatly enlarged by means of a grant from the Ministry of Agriculture and Fisheries.

During 1922 the colony was kept at the Zoological Department of the University of Manchester, but during 1923 it was moved to the Department of Physiology of University College, London, where it is still maintained. On account of the disturbance caused by removal very little breeding was carried on during 1923, but during each of the last two years close on one thousand normal young have been produced. Since the colony is now devoted primarily to experimental work these young represent the stock breeding which is necessary to maintain a supply of animals for experiment. Only the offspring of normal mice, untreated in any way, will be considered in the present paper. The possible factors influencing the sex-ratio to be considered are much the same as those discussed in the previous paper, but the fact that records are now available for four years makes it possible to consider the question of annual variation in the proportion of males. Since the previous papers included a summary of the relevant work of other authors, this aspect will only be briefly mentioned in the present communication.

Of the two previous papers, the first, dealing with the sex-ratio, was confined to the young of ascertained sex born in that year, but the second paper, dealing with fertility included other miscellaneous young of unknown sex*. In the present paper, both fertility and sex-ratio will be dealt with together, and since improvements in technique make it possible to ascertain the sex of all of the individuals of most litters born, the study of fertility will be limited to young of known sex.

The present system of normal breeding is as follows. Twenty specially selected stock bucks, renewed from time to time, are kept as stud animals, and females as available are put into their cages. On the detection of the vaginal plug (indicating copulation) or when pregnancy is obviously established, the female is removed to a separate cage to await parturition. The labour involved by daily examinations for vaginal plugs is a serious drawback, and usually this method is. only used when it is essential that the date of conception should be known. As soon as parturition takes place the young are examined for sex.

In the case of the mouse where the external sex differences are not marked this is far from easy and requires considerable experience. There are however certain features which provide a basis for sexing. Jackson (3) has dealt with these as regards the rat, and much the same kind of observations are necessary in the case of the mouse, except that, owing to its smaller size, the recognition of sex at birth in this animal is even more difficult than in the rat. Macroscopically the external genital organs at birth in the mouse consist merely of the genital tubercle (destined to become the penis of the male or the clitoris of the female). The scrotum is indistinguishable from the perineum, and no sign of the vulva is apparent. The nipple areas in the female are also invisible at birth and in both sexes the urethra runs through the genital tubercle. The early recognition of sex depends therefore on the size, shape and position of the tubercle. In the female this organ is smaller and nearer the anus than in the male. In the female the extremity tends to be indented, while the ventral surface in the case of the male is shiny. These differences make it possible to recognise in most cases the sex of the new-born mouse, but since the differences are purely relative it is often necessary to compare the different individuals of a litter, and in certain cases the two sexes approximate so closely that dissection is desirable if absolute certainty is required immediately. Normally, however, sex can be recognised at birth, and usually the sex of 17-19 day foetuses can be determined by the same signs.

When the young are 9–10 days old the nipple areas in the female can be plainly seen as semi-nude areas in the growing fur, and if this stage is watched for the sex can be recognised with absolute certainty. Subsequently, however, this sign tends to disappear. The vulva and the scrotum can be recognised at about the 20th day.

When the young have been sexed the number of the litter, if large, is reduced to conform with the probable capacity of the doe for suckling. The growth curves of young mice during the suckling period (Parkes (16)) show, naturally, that litters of only one grow faster than any others, and in general that the larger the litter the less rapid the growth of the individuals composing it. The inferior nursing obtained by individuals of litters of 9 and 10 usually retards their growth long after the actual period of suckling. The average growth curves for individuals from litters of 4–7 do not, however, show any very great differences, and since it is impracticable to reduce all litters to one or two, the number (if necessary) is reduced to 4–7 according to the condition of the doe; young, old and inferior does being left with the smaller number. The young removed are occasionally foster-mothered on to does which have just finished suckling their own litter, but the exhaustion of the female prevents the foster-young doing very well, and in any case not more than three young should be fostered in this manner. By far the best way of foster-mothering young is to use a female who has herself just had a small litter, among which one or two extraneous young can easily be mixed without harm. In the rare cases where it is necessary to resort to dissection to ascertain the sex of certain of the young this means can be taken of reducing the size of litter.

The young are weaned at three weeks old, and are reared in special large cages. If, however, the young are taken straight from the mother and put in the large cages a very marked check is sustained and the method now adopted is to wean them by removing the mother from the nest, and to let the young stay in the original nest for a week or 10 days until they become used to the absence of the doe. They are then removed to the large rearing cages. By this means it has been found possible to get the young to grow steadily until they become adult. The sexes are separated at the time of removal from the nest to the rearing cages.

Ovulation usually first takes place during the 8th week, but the young adults are not usually mated until about 10 weeks of age, though this again depends on their physical condition.

The sex-ratio throughout the present paper is calculated as the percentage of males, of which the probable error can be arrived at from the formula ·6745 , where m and f are the percentages of males and females respectively, and n is the number of cases. The probable error of the difference between two male percentages is calculated from , where A and B are the probable errors of the two percentages to be compared.

In this, as in other papers, the term “fertility,” as applied to mice, is used to denote the number of young per litter, and “fecundity” to mean the number of matings which result in offspring. In the present case the latter is measured as the percentage of females in which the detection of the vaginal plug is followed by pregnancy.

The births during 1923–4–5 total 1872, the quota of the respective years being 109, 812, 951. The grand total of mice of known sex produced up to date (including 1922) is thus 2903. The 1872 mice born during the last three years were from 303 litters, giving an average litter size of 6·18. The grand total for the four years 1922-5 is 458 litters, with an average size of 6·34.

The frequency distribution of litter size for the three years under review is as follows :

Table I.

Frequency distribution for size of litter (1923–5).

Frequency distribution for size of litter (1923–5).
Frequency distribution for size of litter (1923–5).
Table II.

Annual variation in sex-ratio and fertility.

Annual variation in sex-ratio and fertility.
Annual variation in sex-ratio and fertility.
Table III.

Comparison of annual ratios.

Comparison of annual ratios.
Comparison of annual ratios.

For the frequency distribution during the combined years 1922–3 to 1924–5 δ2 = 4·46 and δ = 2·11. This standard deviation is markedly less than that of 2·5 previously reported for 1921–2, but it is of the same order as those previously reported for two separate groups of the present material, i.e. δ = 2·10 for 735 young (Parkes (14)) and δ = 2·05 for 407 young (Parkes (16)). The difference of the standard deviation from that for 1921–2 is due to the fact that for the latter year numbers of the larger litters were found, litters of 12, 13, 14 being recorded, and this again is partly due to the fact that the fertility date for 1921–2 were based on the whole of the litters born, while the present data are for litters of known sex only. At the same time, however, the average litter size of 6-18 compares unfavourably with the figure of 6·65 for the 155 sexed litters of 1921–2.

As regards fecundity the data at present are very fragmentary, owing to the labour involved in vaginal examination. So far, however, the history of 88 normal females following the discovery of the vaginal plug has been noted. Of these 71, or a percentage of 8o-8 have proceeded to a normal pregnancy.

The 1872 young born in the 303 litters under discussion comprised 968 males and 904 females, a male percentage of 51·7 ± ·77, which is considerably less than that of 54·2 ± 1·04 for the 1031 young born during 1922. The difference between the two ratios is 2·5, of which the probable error is . The difference is thus nearly twice its error, and tends to be significant. The possible explanation of this difference will be considered later when the question of seasonal variation is considered.

For the 2903 young produced in the four years (1526 males and 1377 females), the male percentage is 52·6 ± ·62.

For the purposes of these experiments the year is taken as running from November to October. This arrangement is convenient for two reasons. October represents the end of the breeding season for mice kept in atmospheric temperature, and, secondly, the move so far experienced by the colony took place at the beginning of the academic year, and any future moves will probably be at a similar time. The mouse breeding records are therefore based on the year November 1st to October 31st. At Manchester University the colony was kept in practically atmospheric temperature and breeding only took place between March and October, 1922, but as the whole available stock was then devoted to normal breeding considerable material (considered as belonging to the year 1921-2) was collected.

Records for four years are thus available for analysis as regards annual variation in sex-ratio and fertility, but for the second of these years (1922–3) only small figures are available, and during this second year, also, the mice were subjected to considerable disturbance. The actual figures are as follows:

As these four years stand it does not seem possible to attach much significance to the variation in the male percentage. The actual figures run as follows:

None of these differences is therefore really significant, and, although the ratio for 1921–2 tends to be significantly different from the rest combined, it is clear that in the colony’s present quarters during the last two years there has been no appreciable variation in the male percentage, and this percentage may apparently be regarded as stabilised so far as annual variation is concerned.

For the year 1921–2 the average size of litter was found to be greatest in the quarter July-September (7·14) and least for January-March (6·03) while April-June and October-December fell in between with a fertility of 6·40 and 6·45 respectively. The births for 1922–5 work out as follows:

These figures suggest that, although the monthly averages for litter size are not very coherent, quarterly variation is found in fertility. Such variation was also found to occur in the data for 1921–2 (all litters) but in this latter case the biggest average size of litter was found in the July-September quarter and the lowest fertility in January-March, while for 1922–5 the highest fertility is in October-December and the lowest in April-June. The meaning of this is not clear but both sets of data agree in giving a higher fertility to the second half of the year than to the first half. This seasonal variation in fertility has one implication of importance. If the total births for a year mainly fall in one quarter, the average size of litter for that year will be largely determined by the figure for that quarter, and if large variation is found from season to season the annual figure may be more representative of one quarter than of the whole year. Thus for a year to be representative it should have approximately equal births in each quarter (a state of affairs not always obtainable in breeding practice) or failing this the average figure for the year may be arrived at by finding the mean of the four quarterly figures. Thus the four quarterly figures for all litters born 1921–2 (Parkes(12)) run 6·03, 6·40, 7·12, 6·45, but since the third quarter with the high fertility provides just half of the litters of the year, the average figure for the year is mainly determined by this quarter and is therefore unduly high at 6·72. The mean of the four quarterly figures is 6·50, a figure more comparable with that of 6-i8 for the last three years. In the latter set of data the average for the year and the mean of the four quarterly values are the same to two places of decimals.

Seasonal variation in the sex-ratio has received much attention. In man it has been both postulated (Heaped)) and denied (Bonnier (1)). In the pig it has not been found (Parkes (10); Machens (9)), while for the white rat its occurrence has been demonstrated by King and Stotsenburg (6). In Peromyscus Sumner found definite monthly variation which could only be accounted for by some effect of season, but the variation did not occur at the same time as in King’s rats.

In my own mice for the year 1921–2 it was shown that the three months April-June had a significantly lower proportion of males than the months July-October, these being the only months of that year during which appreciable breeding took place. This result agrees fairly well with King’s findings for rats. The seasonal data for the combined years 1922–5 are as follows:

The relations of the ratios for these four quarters to the ratio for the whole of the births are shown in Table VI.

Table IV.

Seasonal fertility, 1922–5.

Seasonal fertility, 1922–5.
Seasonal fertility, 1922–5.
Table V.

Monthly births, 1922–5.

Monthly births, 1922–5.
Monthly births, 1922–5.
Table VI.

Relations of quarterly ratios to the mean ratio.

Relations of quarterly ratios to the mean ratio.
Relations of quarterly ratios to the mean ratio.
Table VII.

Inter-relation of quarterly ratios.

Inter-relation of quarterly ratios.
Inter-relation of quarterly ratios.

From this table it is clear that the second quarter (January-March) and the fourth quarter (July-September) have ratios which are almost identical with the mean ratio. The ratio for the October-December quarter on the other hand is appreciably higher than the mean, while the ratio for April-June is appreciably lower. This result is very similar to that previously reported for 1921–2. In that year the April-May-June ratio was 49·8, while the male percentage for July-September was 56·2, though as breeding then stopped almost completely it was impossible to obtain ratios for the other quarters.

This seasonal variation in the ratio has one implication of great importance. If wide divergences are found in the quarterly ratios it is clear that the relative number of births in each quarter will have an influence on the yearly ratio; the quarter contributing the most births will have a predominating influence on the average ratio. This means that the yearly average may be merely an arithmetic result, and for this reason it may be inadvisable to compare the quarterly ratio with the ratio for the year. In the following table, therefore, the four quarterly ratios are compared with each other.

This table emphasises the fact that practically no difference is to be found between the ratios for the second (January-March) and fourth (July-September) quarters, and the ratios for these two quarters may conveniently be considered as forming a sort of zero-line to the seasonal variation. From this line an upward variation occurs in the first quarter and a downward one in the third and both of these variations tending to be statistically significant (d/e varies between 1·39 and 2·10) from the “zero-line” ratios. When these two variations are themselves compared the difference is found to be very appreciable, and it is, in fact, strongly significant (d/e = 3·31). The curves for seasonal variation in the sex-ratio and in fertility are shown in the following diagram.

In view of this marked seasonal variation in the sex-ratio it would, in order to obtain a valid figure for the whole year, be necessary that fairly equal numbers of births should occur in each season, and this, of course, is not normally convenient or possible in a colony maintained primarily for recruiting experimental stock. The same effect however, can be produced by calculating the figure for the whole year as the mean of the four quarterly figures. Thus, in the present instance, the ratio of the 1872 births for the three years is 51·7, whereas the mean of the four seasonal ratios (55·9, 51·9, 48·2 and 52·2) is 52·0. In this case no great difference exists between the two figures, but this is due solely to the fact that the seasonal births are fairly equal and before one year can be definitely said to have a different sex-ratio from another it would appear to be desirable to calculate the figure for the year according to both these methods. A further implication of seasonal variation is that unless a breeding experiment lasts over a whole year, the control figure used for the sex-ratio should be that for normal mice bred in the same season, and not merely an annual figure.

The results of other authors on the influence of litter size was briefly summarised in the previous paper dealing with my records for 1921–2, and this paper may be consulted for references. The whole consensus of opinion is that no definite correlation exists between the size of litter and the proportion of males. The conclusion was also arrived at by Sumner, McDaniel and Huestis(17). My own records for 1921–2 seemed to suggest that small litters (14) had a sex-ratio below the normal, but the further records given below for the last three years do not confirm this conclusion (Table VIII).

Table VIII.

Male percentage according to litter size, 1922–5.

Male percentage according to litter size, 1922–5.
Male percentage according to litter size, 1922–5.

The grouping employed in this table is, of course, quite arbitrary, but any grouping on similar lines would give similar results. The relations of the group ratios to each other and to the average for all litters is given in Table IX.

Table IX.

Relations of litter size group ratios to each other and to the mean.

Relations of litter size group ratios to each other and to the mean.
Relations of litter size group ratios to each other and to the mean.

It is clear that none of these group ratios shows any significant variation from each other, or from the average ratio for all litters. The d/e value for group 5–7 and 8–11 is more noticeable than the others, but even so it hardly approaches real significance, and in any case the variation is in the opposite direction to that found for 1921–2. The tentative suggestion of positive correlation between proportions of males and litter size cannot therefore be substantiated, and must presumably be abandoned, so far, at any rate, as the records for 1922–5 are concerned. The present work is, therefore, brought into line with these authors who have failed to demonstrate any definite regular condition between litter size and sex-ratio.

It does not seem possible to add a great deal to the consideration of the factors governing the sex-ratio in mice which was made in the earlier paper. It was then attempted to show that the variations in the sex-ratio at birth were due either to variation in the ratio at conception or to variation in the amount of sexually different pre-natal mortality. The hypothesis that pre-natal mortality in the mouse falls preponderating on the male was inferred from a previous paper (10), thus making it possible to suppose that decrease of the ratio at birth might be due to increase of pre-natal death. During the last year two papers have been published by Macdowell and Lord (7,8). These authors based their work on the following considerations :

  1. That if pre-natal mortality falls preponderatingly upon the male, the ratio at birth should be inversely proportional to the amount of pre-natal mortality as measured by corpora lutea counts and

  2. That litters which have been subjected to no pre-natal mortality should have a sex-ratio corresponding to the sex-ratio at conception.

Their actual work appears to show that there is no correlation between the amount of pre-natal mortality and the sex-ratio at birth, and that the primary ratio at conception is almost identical with the ratio at birth. Two criticisms can however be urged against these views. In the first place the group with the greatest mortality contained 54 young, of which 21 were males and 33 were females, i.e. 63·6 males for 100 females, and leaving the ratio in that form they say that it is not significant. The male percentage in this group works out at 38-9, and the probable error of this percentage calculated by the normal statistical method is ± 4·47. The other groups together total 1248 males and 1923 females, a male percentage of 50·5 ± ·67. The difference between the two is thus 11·15 ± 4·64, and it seems premature to state that this is quite without significance.

Secondly, as regards the assumption that the ratio for litters which have been subjected to no pre-natal mortality represents the normal primary ratio, it may be urged that there is no reason to suppose that these litters represent a random sample of all litters conceived. If the male embryo were less viable than the female, then litters containing a large proportion of females, or consisting entirely of females, would have a better chance of surviving gestation without loss than would a litter consisting mostly of males. The result of this would be that the complete litters would consist of selected material, i.e. material containing a higher percentage of females than the average at conception.

Nevertheless, though this criticism undoubtedly has weight, it is clear that the work of MacDowell and Lord is both of great interest and importance, and is founded on a technique which holds great possibilities. It would be of much interest if the mouse was different from the rat (King(s)), the pig (Parkes(15)), the cow (Jewell(4)), and the human subject, in having a pre-natal mortality which fell equally upon the two sexes.

As regards the results recorded in the present paper little good can be achieved by further discussion in the present stage of our knowledge. The correlation, shown in Fig. 1, between the seasonal variation in the sex-ratio and in the fertility is suggestive. Since the largest litters occur when the sex-ratio is highest, and the smallest when it is lowest, it might be supposed that this was connected with a differential pre-natal mortality. If however, pre-natal mortality is the chief factor in moulding the size of a litter (as required by this supposition) and is also differential between the sexes, some effect of this ought to be found in considering the relation between litter-size and sex-ratio, whereas, as pointed out above, no direct correlation seems to exist between litter-size and sex-ratio.

Fig. 1.

Seasonal variation in sex-ratio and fertility.

Fig. 1.

Seasonal variation in sex-ratio and fertility.

It is, in fact, quite obvious that a great deal more information is required before any coherent story of the factors governing the sex-ratio at birth in the mouse (and all other mammals) can be put together.

(1)
BONNIER
(
1923
).
Zeit.für Ind. Abs. und Vererb
.
32
.
(2)
HEAPE
(
1909
). “
The Proportion of the Sexes produced by White and Coloured Peoples in Cuba
.”
Trans. R. S. B
,
200
.
(3)
JACKSON
(
1912
). “
On the Recognition of Sex through External Characteristics in the Young Rat
.”
Biol. Bull
.
33
.
(4)
JEWELL
(
1921
). “
Sex-ratios in Foetal Cattle
.”
Biol. Bull
.
41
.
(5)
KING
(
1921
). “
Birth mortality in the Rat and in Man
.”
Anat. Rec
.
20
.
(6)
KING and STOTSENBURG
(
1915
). “
On the Normal Sex Ratio and the Size of the Litter in the Albino Rat (Mus norvegicus albinus)
.”
Anat. Rec
.
11
.
(7)
Macdowell
,
Carleton
and
LORD
(
1925
). “
Data on the Primary Sex-ratio in the Mouse
.”
Anat. Rec
.
31
.
(8)
MACDOWELL and LORD
(
1925
). “
The Sex-ratio in Litters of Mice classified by the Total Amount of Pre-natal Mortality
.”
Proc. Soc. Exp. Biol. Med
.
22
.
(9)
MACHENS
(
1915
). “
Fruchtbarkeit und Geschlechtsverhältnis beim veredelten Landschwein
.”
Berliner Tierärztliche Wochenschrift
.
31
.
(10)
PARKES
(
1923
). “
Studies on the Sex-ratio and Related Phenomena, (3) Influence of Size of Litter
.”
Ann. App. Biol
.
10
.
(11)
PARKES
(
1924
). “
Studies on the Sex-ratio and Related Phenomena, (5) The Sex-ratio in Mice and its Variations
.”
Brit. Joum. Exp. Biol
.
1
.
(12)
PARKES
(
1924
). “
Fertility in Mice
.”
Ibid
.
2
.
(13)
PARKES
(
1924
). “
Studies on the Sex-ratio and Related Phenomena, (1) Fœtal Retrogression in Mice
.”
Proc. Roy. Soc. B
,
95
.
(14)
PARKES
(
1925
). “
The Effects on Fertility and the Sex-ratio of Sub-sterility Exposures to X-rays
.”
Proc. Roy. Soc. B
,
98
.
(15)
PARKES
(
1925
). “
Studies on the Sex-ratio and Related Phenomena, (7) The foetal sex-ratio in the Pig
.”
Journ. Agri. Sci
.
15
.
(16)
“Growth of Young Mice according to Size of Litter.” (In Press.)
(17)
SUMNER, MCDANIEL and HUESTIS
(
1922
). “
A Study of Influences which may affect the Sex-ratio of the Deer Mouse (Peromyscus)
.”
Biol. Bull
.
43
.
*

Where one or more of a litter was eaten before being sexed the litter was discarded from the sex-ratio records, though it was of course available for fertility records.