A non-transmissible plasmid, pAV5, was isolated from a hospital strain of Acinetobacter calcoaceticus, JC17. pAV5 confers resistance to the antibiotics tetracycline and neomycin in the genetically characterized strain of A. calcoaceticus EBF 65/65. Transfer of pAV5 can be mediated by the sex factors pAV1 and R751; transfer is occasionally associated with the segregation of the resistance determinants amongst the transconjugants. Phenotypic dissociation of pAV5 corresponds with the formation of two independent plasmids designated pAV51 and pAV52 mediating resistance to neomycin and tetracycline respectively.

Variable frequencies of unhatched eggs were observed to be produced by a number of F1 interstrain hybrids. This type of partial sterility resulting from F2 embryo death was found to be associated with the P-M system of hybrid dysgenesis. Dysgenic hybrid progeny of crosses between M strain females and P strain males may therefore have reduced fertility due to the disruption of development at two different stages: early F1 gonadal development and early F2 embryo development. These disruptions result in the previously described F1 gonadal dysgenesis (GD sterility) and F2 embryo lethality (EL sterility) respectively. The two morphologically distinct types of P-M-associated sterility differ in their patterns of response to F1 developmental temperature, and the temperature-sensitive period for EL sterility occurs considerably later in F1 development than for GD sterility. EL sterility is similar to SF sterility, which is associated with the I–R system of hybrid dysgenesis in that both result from death during early F2 embryogenesis. However, EL sterility differs from SF sterility in not being restricted to hybrids of the female sex and in showing different patterns of response to temperature and ageing in the F1 generation. Some implications of the existence of EL sterility for methods of strain classification in the I–R system are explored.

The protein and amino acid compositions of hair from mice carrying the naked (N) gene were compared with those of wild-type (+ / +) mouse hair using two-dimensional polyacrylamide gel electrophoresis and amino acid analysis. In samples obtained from N /+ mice, the numbers of positions of the low-sulphur (LS), high-sulphur (HS) and high tyrosine (HT) protein spots were indistinguishable from those observed in + / + samples but the amount of HT protein was reduced in N / + hair. In samples obtained from N /N mice the protein spots were fewer than, and perhaps different from, those in samples from the other two genotypes; all protein classes were represented but the HS and HT protein content appeared to be less than that in + / + and N / + hair. Amino acid analyses of hair samples showed that the HS contents of + / + and N / + hair were approximately the same, whereas the HT protein content of N / + hair was about half that in + / + hair. N / N hair contained less half-cystine and tyrosine than + / + and N / + hairs, but because the contents of proline, threonine and glycine were unexpectedly high, it appears that protein components of unusual composition are present in N / N hair. It was concluded that the gene at the N locus is involved only indirectly in the synthesis of the structural proteins found in mouse hair.

Thirty-two t haplotypes were extracted from wild mice captured in Central Europe, Spain, the Soviet Union, Israel, Egypt, the Orkneys and South and North America, and tested for lethality in the homozygous state. Twenty-two proved to be homozygous lethals, 8 semilethals and 2 viables. The lethal t haplotypes were then tested by the genetic complementation test for identity with representatives of known complementation groups and with each other. Five of the 22 haplotypes proved to carry previously identified lethality factors (tw5, tw73, and tLub-1), while the rest carried new factors. The 17 haplotypes fell into 8 new complementation groups. Two of the new groups are partially overlapping in that they seem to share some lethality factors and differ in others. These tests raise the total number of known complementation groups to 16. The distribution of the individual t haplotypes among wild mice populations seems to reflect their differentiation from a common ancestor haplotype.

Puberty was studied in mice of the ninth selected generation of the Q-strain. There were 6 replicate lines selected for large body size (6-week weight), 6 replicates selected for small size and 6 replicate unselected controls. Female puberty was assessed by the opening of the vagina and male puberty by the first copulation plug. The sexes differed in the mean age at puberty, males being older by 13 days in the large, 4 days in the control and 8 days in the small lines. The sexes differed also in the way size affected puberty. In males the large and small lines reached puberty at the same age and both were older than the controls. In females the large lines on average were heavier and younger at puberty than the controls, and the small lines were lighter and older than the controls, though not significantly older. The replicates within each size-group, however, reached puberty at about the same weight, irrespective of their differences in growth rate. Thus, the differences of growth between the large, control and small groups affected both the weight and the age of females at puberty, but the differences of growth between the replicate lines within each size affected only the age at puberty. No explanation was found for this inconsistency between size-groups and replicates. Several lines of evidence led to the conclusion that in females puberty is partly or mainly weight-dependent, whereas in males it is almost wholly age-dependent.

Female reproductive performance is reported in mice selected for ten generations for one of three criteria: either appetite (A), fat percentage (F) or total lean mass (P). For each criterion lines were selected for high (H) or low (L) performance, with contemporary unselected controls (C). In the A and P lines, litter size changed in the direction of the selected criterion, the changes being larger and more rapidly established in the A than in the P lines. At generation 10, the differences in litter size between high and low lines were 2·6 live young born in the A lines, and 1·0 live young born in the P lines. The differences in 6-week weight between the high and low lines were 3·5 g in the A lines, 6·5 g in the P lines. Changes in ovulation rate were the primary reason for changes in litter size, the differences between the high and low lines being 3·8 corpora lutea for the A lines, and 3·1 corpora lutea for the P lines. Fitting body weight at mating as a covariate within lines in the analysis of ovulation rate and live foetus number removed the differences between the high and low selected P lines, but not those in the A lines. The high and low selected A and P lines did not differ in prenatal survival. There were no consistent differences in litter size, ovulation rate or pre-natal survival in the F lines.

A method for analysing probabilistic models representing an extension of path analysis, and a model of mixed homogamy based on this method were introduced recently (Rao, Morton & Cloninger, 1979; Cloninger, 1980). However, constraints imposed by the method on linear models, or a rationale for and implications of the mixed homogamy model have not been clearly stated.

A mathematical treatment of the extension of path analysis and of the mixed homogamy model is presented in this paper. Constraints on linear models imposed by the extension are obtained. It is demonstrated that the mixed homogamy model, when applied to analysing nuclear families in a population, implies the following mating system in the population: some individuals choose their mates strictly on the basis of group membership, others choose their mates strictly on the basis of phenotype, and no individual chooses a mate on the basis of both group membership and phenotype.

Multilocus recombination frequencies are expressed in terms of pairwise recombination frequencies for paracentric and pericentric intervals. Applications of this theory to mapping and genetic counselling are discussed.

A single locus model of extranuclear genomes is developed under the assumption of the complete action of within-generation drift which is caused by random transmission of multiple copy genomes during cell division in a generation. Within-generation drift segregates different copy genomes in a cell into different cells, resulting in homoplasmic cells. Under some conditions, the present model reduces to that for haploid nuclear genomes. A point overlooked in previous models is that the multiplicity also admits of the possibility of selection occurring within a cell or between cells in an individual (within-generation selection). If there is selection mediated by, for instance, differential proliferation of genomes, then a haploid model no longer explains the dynamics of extranuclear genomes. Rather a model analogous to biased gene conversion at a single locus (Nagylaki, 1983; Walsh, 1983) is more appropriate. An application of this model to either the fixation probability or substitution rate of new mutations shows that strictly maternal inheritance does not allow the fullest use of mutations, as it obscures the effect of within-generation selection. But if there is appreciable paternal contribution, within-generation selection could be a strong evolutionary force to which nuclear genomes are never exposed.

In Takahata & Slatkin (Genetical Research 1983, 42, 257–265), two sentences should be read as:

but, in contrast, the configuration of the genome and the mode of transmission both make a large difference (on page 257, lines 11–12 up)

Contribution no. 1528 from the National Institute of Genetics, Mishima, Shizuokaken 411, Japan (on page 264, above REFERENCES).