I am very honoured to be invited to give this Lecture. I knew John Hammond well, and am personally in his debt for the many kindnesses and considerations he extended to me early on in my career. He was, in more ways than one, a giant of a man, and it is most fitting that this Society should honour his memory by inviting special lectures in his honour, and especially on the 50th anniversary of the founding of the British Society of Animal Production (BSAP) in 1944.

The present study was undertaken to assist conservation and improvement schemes in the Sahiwal breed of cattle in Pakistan. A data set, consisting of records of 244 pure Sahiwal breeding bulls and 5247 cows, the latter representing about 80% of all recorded Sahiwal cows in Pakistan born during a period covering about 20 years, was analysed with regard to inbreeding, additive relationships, effective population size and generation intervals. Average inbreeding coefficients of 1224 cows and 49 bulls, for which at least the grandparents and great-grandsires were known, were 0·043 and 0·046, respectively. About two-thirds of the inbreeding was due to matings between animals with parents or grandparents in common. The mean additive relationship among the cows was 0·062, with within-herd averages ranging from 0·087 to 0·358. The average population size in a subdata set of recorded Sahiwal cattle from 1980 to 1984 was 1612, whereas the most likely estimate of the effective population size was about 30 animals for the same active breeding population. The study indicated the immediate need for an active conservation programme whereby the Sahiwal subpopulations of India and Kenya also should be involved.

Twelve lactatiug dairy cows were offered either fresh grass (G) or grass silage prepared from the same pasture which had undergone either an extensive (E) or restricted (R) fermentation, in a three-period change-over design experiment. Ensilage resulted in a reduction in forage pH and water-soluble carbohydrate concentration and an increase in ammonia-nitrogen concentration. The ensiling techniques used ensured that both silages were well preserved with pH values for E and R of 3.81 and 4.08 respectively, while the respective lactic acid concentrations were 124.6 and 27.0 g/kg dry matter respectively. Whilst there were no significant differences in dry-matter intake between treatments, when corrected for losses of volatile compounds, animals offered silage with a restricted fermentation consumed their food at a higher rate of intake (P < 0.001) than did animals on the other treatments. It was also found that while there was no significant difference in milk yield between animals offered fresh or ensiled forage, both butterfat and protein concentration and yields of butterfat were significantly higher (P < 0.01 or greater) with grass and restricted fermented silage than with extensively fermented silage. Ensiling of herbage had no significant effect on apparent digestibility. It is concluded that ensiling of herbage per se had little effect on overall animal performance but pattern of silage fermentation resulted in alterations in milk composition.

A study was carried out to examine the effects of ensiling and pattern of silage fermentation on nitrogen and energy utilization, rumen fermentation and rumen degradation characteristics of forage by lactating dairy cows. Six nonfistulated dairy cows and three rumen fistulated dairy cows were offered fresh grass (G) or grass silage produced from the same sward which had undergone either an extensive (E) or restricted (R) fermentation. Animals offered extensively fermented silage had liigher urinary nitrogen (N) outputs (expressed as a proportion of N intake) than those offered fresh grass (P < 0·05). Corresponding values for animals offered restricted fermented silage were intermediate between those offered grass and extensively fermented silage. Animals offered fresh grass and restricted fermented silage also had higher methane energy losses (P < 0·05), than those offered extensively fermented silage. The ensiling of grass had little effect on metabolizable energy (ME) intake or on the efficiency of utilization of ME for lactation (k1). Animals offered extensively fermented silage had higher proportions of propionate and lower proportions of acetate than those offered fresh grass or restricted fermented silage (P < 0·001). Ensiling resulted in an increase in the soluble (a) fraction and a reduction in the potentially degradable (b) fraction of N.

The study determined the effects of partial removal of alkaloids (detoxification) in crushed lupin seed (CLS) on voluntary food intake, and yield and composition of milk from dairy cattle. Twenty multiparous Friesian dairy cows (first 90 days of lactation) were assigned, according to a randomized complete-block design, to five diets. The diets were formulated to be isonitrogenous (25·6 g N per kg of diet dry matter (DM)) and contained napier grass, lucerne hay, maize bran and urea. The control diet (CON) contained sunflower meal; two diets contained intact CLS at 150 (LUI-15) or 300 g (LUl-30) per kg diet DM. The other two diets contained detoxified CLS at 150 (LUD-15) or 300 g (LUD-30) per kg diet DM. Lupin seeds were detoxified by treatment with boiling water, followed by steeping in cold water. The diets were analysed by gas chromatography for the alkaloids, lupanine and 13-hydroxylupanine. The total alkaloid content of LUI-15 and LUl-30 was 3·8 and 8·0 g/kg diet DM, respectively; by contrast that of LUD-15 and LUD-30 was 2·1 and 5·2 g/kg diet DM respectively. Increasing the level of intact CLS in the diet led to a decrease in voluntary food intake. Cows given LUl-30 had a lower milk yield (11·1 kg/day) than those given LUI-15 (13·8 kg/day; P < 0·01) but there was no difference in either milk protein yield or content. Detoxification of lupin removed proportionately about 0·40 of the total alkaloids from intact CLS and increased food intake, and the yield of milk and milk protein (P < 0·05) but reduced milk fat content. Detoxification of CLS also reduced the rumen degradability of lupin protein (P< 0·05). It is concluded that the reduction in organic matter intake and milk yield of cows given diets containing intact CLS was due to the presence of lupanine and 13-hydroxylupanine. To maximize its usage in diets for dairy cattle, lupin should be detoxified; it can then be included at levels up to 300 g/kg diet DM.

An experiment was carried out to evaluate dairy performance and rumen fermentation from a forage-based diet supplemented with Lucaena leococephala (leucaena) and maize bran. Bos indicus (Ayrshire/Brown Swiss) х Bos taurus (Sahiwal) cattle were offered Pennisetum purpureum (napier grass) ad libitum either unsupplemented (N) or supplemented on a dry-matter (DM) basis with 1 kg (LL) or 2 kg leucaena (LH) or 2 kg leucaena together with 1 kg DM maize bran (LHM). Food intake, milk yield, diet apparent digestibility and nitrogen balance were measured using eight cows (mean live weight 384 (s.d. 41) kg in two 4х4 Latin squares. Rumen fermentation was measured using four rumen fistulated steers (mean live weight 352 (s.d. 12) kg in a 4 х 4 Latin square. Performance was recorded during the last 7 days of each 21-day experimental period. Supplementation increased food intake (6·3, 7·6, 7·9 and 8·7 kg DM per day; P < 0·01; s.e.d. 0·24) and milk yield (5·1, 5·4, 5·5 and 6·5 kg/day; P < 0·01; s.e.d. 0·28) for N, LL, LH and LHM respectively. Supplemented diets tended to be more digestible compared with napier grass as the sole food (organic matter apparent digestibility coefficients 0·54, 0·55, 0·57 and 0·57; s.e.d. 0·020; N, LL, LH and LHM, respectively) and N balance increased progressively with supplementation.

Supplementation progressively increased rumen ammonia-nitrogen concentrations in steers and the rate of rumen degradation of napier grass increased when leucaena (diets LL and LH) was offered. The results suggest that supplementation of napier grass with leucaena increased DM intake owing to increased ammonia concentrations and rates of napier degradation in the rumen. Supplementation of the forage diet with 1 kg DM maize bran did not significantly reduce the rumen degradation of napier grass and increased the molar proportions of propionate which may have increased the efficiency of utilization of metabolizable energy and therefore milk production.

Forty pregnant F1 crossbred dairy cows (20 Friesian × Boran and 20 Simmental × Boran) were stratified in a 2 × 2 diet × work factorial experiment (not working-not supplemented, NWNS; not working-supplemented, NWS; working-not supplemented, WNS; and working-supplemented, WS). Working cows pulled sledges 100 days/year (pull = 350 to 450 N, 4 h/day, 4 days/week). Work output of supplemented and non-supplemented cows was similar over 1 and 2 years. Over all 3 years, dry-matter intake relative to metabolic body size (g/kg M0.75 was greater for working, compared with non-working cows. Body-weight changes and body condition score were similar for working and non-working cows. Non-supplemented cows lost weight throughout the first 2-year period, while supplemented cows tended to maintain or gain body weight over 1 and 3 years. Over 2 years, supplementation of working cows proportionately reduced live-weight loss by 0.73 and doubled the number of conceptions and parturitions. Days in milk, milk, milk fat and protein yields were similar for working and non-working cows, but were greater for supplemented, compared with non-supplemented, cows. Total conceptions and calves born in all 3 years tended to be greater for supplemented and non-working compared with non-supplemented and working cows. A productivity index (PI) that took into account food intake was calculated. The PI for supplemented cows over 2 years was greater than that for non-supplemented cows. Meanwhile the PI was similar for working and non-working cows over all periods considered. A similar PI for working and non-working cows under supplementation indicates potential of on-farm adoption of a cow traction technology that includes improved food production and ng strategies.

The synchronization of eating, ruminating and idling activity by lactating dairy cows grazing a perennial ryegrass-white clover sward was studied. Synchronization was defined as the number of pairs of cows engaged in a particular activity as a proportion of the total possible number of pairs and was compared with random expectation using a kappa statistic. All three activities were significantly more synchronized than random expectation. This suggests that wherever possible individual cows should not be treated as replicates in grazing experiments.

In a prospective 2-year study, milk progesterone and oestrone sulphate determinations were used to monitor and assess pregnancy in lactating cows. The testing was done centrally at a government regional veterinary establishment which was responsible for providing semen to the farms of the area and also had available specialist expertise in animal reproduction. Sample collection and dissemination of test results were achieved through the network of the artificial insemination service. This approach was preferred because it was considered ergonomic, cost-effective and reliable. Comparison of oestrus behavioural signs and progesterone on the day of insemination revealed that 7·8% (89/1133) of the animals tested had high progesterone levels. Single progesterone testing on day 21 post insemination was not satisfactory for pregnancy diagnosis as the positive predictive value was 83·0% (147/177). The combination of low progesterone on day 0 and high progesterone on day 21 improved the positive predictive value to 87·4% (235/269). Results from 400 samples taken 110 to 130 days post insemination from 53 animals revealed that cows with oestrone sulphate concentrations greater than 0·1 μg/l actually calved. This assay appears to be particularly suitable as a pregnancy confirmation test. A pilot field study implementing a programme of testing on days 0, 21 and 42 for progesterone and between days 110 to 130 for oestrone sulphate not only achieved almost absolute pregnancy predictive values but careful result analysis strongly indicated that the approach of systematic testing can also reveal the fertility profile of a given herd and help to identify potential causes of pregnancy failure from the period of the reproduction cycle in which they occur.

Sixteen Finnish Ayrshire cows were used in a four period cyclic change-over experiment to evaluate eight concentrate supplements in a 2 × 2 × 2 factorial arrangement. Two concentrate energy sources (starchy and fibrous) were used, each given with two levels (0 and 180 g/kg on a dry-matter (DM) basis) ofrapeseed meal (RSM) and two levels (0 and 180g/kg on DM basis) of wet distillers' solubles (WDS). The starchy concentrate (S) comprised rolled barley and oats (1:1). For the cows given the fibrous concentrate (F), 600 g/kg of the grain mixture was replaced with a mixture of fibrous by-products. Grass silage was offered ad libitum and the supplements offered at a rate of 9 kg/day (fresh weight).

The cows offered F supplements consumed more silage DM than those offered S supplements (P < 0·01) but the difference in the total DM intake was smaller. The cows given S supplements produced slightly more milk than those given F supplement (26·5 v. 25·9 kg/day) but no differences were observed in the yield of milk constituents between the energy supplements. Replacing starch with fibre in the concentrate had no effect on the apparent digestibility of organic matter but tended to increase that of neutral-detergent fibre.

Including RSM in the supplement increased total DM intake (P<0·05) and led to increases in the yield of milk and milk constituents (P < 0·001 and P < 0·01). Diet apparent digestibility was not affected by RSM supplementation.

There were no significant interactions between concentrate energy source and RSM supplementation. Including WDS in the diet had no effect on food intake. Production responses to WDS supplementation varied with the concentrate energy source. WDS increased (at least P< 0·05) milk yield (1·0 kg/day) and the yields of fat (76 g/day) and protein (48 g/day) when given with the starchy concentrate. With the fibrous concentrate WDS had no effect on milk and protein yield, and decreased fat yield by 52 g/day (P < 0·05). The positive responses in the yields of milk fat and protein to RSM and WDS supplementation were almost additive with the starchy concentrate. Inclusion of WDS in the diet increased the apparent digestibility of organic matter and gross energy.

Forty pregnant F1 crossbred dairy cows (20 Friesian × Boran and 20 Simmental × Boran) were stratified in a 2 × 2 diet × work factorial arrangement (not working-not supplemented, NWNS; not working-supplemented, NWS; working-not supplemented, WNS; and working-supplemented, WS) were used to study the effect of work and diet supplementation on roughage intake, apparent digestibility of nutrients, digesta kinetics and plasma metabolites. Dry matter (DM) and organic matter (OM) intakes were greater for working than for non-working cows. Work and supplementation increased DM and OM in vivo apparent digestibility. This effect was particularly significant when comparing working cows at rest and at work. Passage rate and rumen mean retention of Cr-mordanted hay were similar for working and for nonworking cows. Liquid turn-over rate of Co-ethylene diamine tetra acetic acid (EDTA) was lower and rumen mean retention time of Co-EDTA was greater in working than in non-working cows whether supplemented or not. Working coios had lower plasma glucose during and immediately after working than non-working cows. Plasma non-esterified fatty acid (NEFA) was higher in working cows at the end of the working period and after rest between working hours. This effect was greater in non-supplemented than in supplemented cows. In working cows, β-hydroxybutyrate was lower at the end of working hours than in non-working cows. Lactate was greater in working than in non-working cows after the 3rd and 4th hours of work. Digesta kinetics could explain only partially the possible mechanisms responsible for greater roughage intake and apparent digestibility in working cows. The decrease of plasma glucose and the increase in NEFA during work indicates an increasing utilization of NEFA by muscle during consecutive working hours.

Heritabilities (h2) and genetic correlations (rc) involving body weights, measures of testes size and ewe lamb reproduction were calculated using 953 spring-born lambs produced during formation of a three-breed composite population containing 0·50 Dorset, 0·25 Rambouillet and 0·25 Finnish Landrace breeding. The h2 of body weight ranged from 0·2 to 0·3 between 45 and 150 days of age, but increased to 0·41 at the start of breeding of ewe lambs at about 6 months of age. Additive maternal effects were not important, perhaps because intensive housing encouraged cross-suckling. Testes size was measured as scrotal circumference (SC), SC with linear adjustment for weight (SCW) and SC divided by the one-third power of body weight (RSC). The h2 of SC traits was highest at 90 days (0·52 for SC, 0·62 for RSC and 0·57 for SCW), and h2 for age at maximum SC growth rate (i.e. TI, the inflexion point of a logistic SC growth curve) was 0·49. The h2 of ewe lamb fertility in autumn was 0·09 but the h2 for frequency of remating the following spring was 0·41. The h2 for ewe lamb litter size in these data was zero. At 90 days, rc between alternative measures of testes size exceeded 0·86; rG with body weight were 0·67 for SC, 0·53 for RSC and –0·74 for TI. The rG with ewe lamb fertility for 90-day weight, SC, RSC and SCW and for TI were –0·25, 0·20, 0·25, 0·38 and –0·32, respectively. The rc for these traits with spring remating were 0·34, 0·34, 0·26, 0·10 and –0·48, respectively.

A nucleus dairy population using multiple ovulation and embryo transfer (MOET) was stochastically modelled with overlapping generations. The aim was to investigate the feasibility of controlling inbreeding in MOET breeding schemes using more realistic parameters for embryo recovery and best linear unbiased prediction (BLUP) for genetic evaluation. Four different cases (involving the culling of donors, more donors and the use of organized progeny testing of nucleus bulls) were studied in combination with nested and factorial designs. Further studies involved modifications of the selection index, including subtracting parental breeding values, inflating the genetic variance in the BLUP evaluation and penalizing inbred animals; these options were examined both with and without organized progeny testing. The effects of applying these schemes on both genetic response and rate of inbreeding were investigated. The results stressed the importance of incorporating progeny testing into MOET schemes for value of reducing inbreeding whilst maintaining genetic progress. There was no significant difference between nested and factorial designs. In the absence of progeny testing the inflation of genetic variance was more effective than subtracting parental breeding values at controlling inbreeding; however incorporating progeny testing made the latter strategy more potent and the superiority of inflating the genetic variance was in this case much smaller and non-significant.

The commercial value of animal carcasses depends not only on their weight but also on their composition and shape (termed conformation). This is usually assessed subjectively by a skilled inspector. In this paper an attempt is described to assess the saleable meat yield of sheep carcasses by automatic digital image analysis. A low-cost system based on a still video camera and a personal computer was used. The results indicate that better prediction of saleable meat yield can be obtained using objective measures of carcass shape than from subjective conformation scores. Information from the intensities of colour components was not found to be useful, possibly due to difficulties with lighting and image quality. Recommendations are made for implementing a practical system.

Weaning weights (or weight nearest to 205 days of age) from 8086 Mashona calves were collected from seven herds covering the period 1976 to 1988. Estimates of (co)variance components were obtained by restricted maximum likelihood using a derivative free algorithm and fitting an individual animal model. Estimates of direct additive heritability, maternal additive heritability, their correlation, total heritability and repeatability, obtained under two models were 0·243 and 0·281, 0·113 and 0·392, –0·282 and –0·269, 0·252 and 0·298, and 0·409 and 0·573, respectively. Trends within herd were estimated from the mean value of progeny born within a particular year. There were no significant trends in direct additive breeding values. A general decline in maternal breeding values was observed. Only one herd (herd 1) had a significant eroiyonmeYital freni (0·385 kg/year). It IMS emphftsilfiti that any future revision of the method used to improve the weaning weights of beef cattle should also improve their maternal breeding values.

The effect of bromocryptine on milk production was investigated in 32 crossbred ewes (Flemish Milksheep х Suffolk) with one sucking lamb, during lactation and during drying off. After weaning the lambs, all ewes were hand milked twice daily for 3 days. From day 4 onwards, the ewes were divided into four groups with equal mean daily milk production. Ewes of groups A and B were hand milked twice daily during the following 9 days. Ewes of groups C and D were dried off by reducing the milking frequency. They were milked on days 5, 8 and 12. In groups B and D, bromocryptin (1 ml of 1 g/l ethanol 50%) was injected intramuscularly twice daily while control ewes (groups A and C) received 1 ml ethanol 50%. Blood samples were taken by venipuncture prior to each injection and were assayed for prolactin, growth hormone, thyrotropin and cortisol.

Bromocryptine decreased plasma prolactin levels but did not alter growth hormone or cortisol concentrations during lactation or drying off. Thyrotropin levels were elevated by bromocryptine during drying off but were unaffected during galactopoiesis.

During lactation the daily milk production remained unchanged in control ewes while in bromocryptine-treated ewes a proportional decrease of 0·5 was observed. In all ewes the decrease in milk production occurring at drying off could not be accelerated or stimulated by bromocryptine.

It is concluded that prolactin plays an important role in galactopoiesis in ewes but its role is comparatively unimportant compared with other factors during mammary involution.

Four diets, barley/proprietary protein concentrate (B/PC), barley/maize gluten (B/MG), barley/rapeseed meal (B/RSM) and wheat/'rapeseed meal (W/RSM) were formulated to contain 165 g crude protein and 13·0 MJ metabolizable energy per kg dry matter. In experiment 1, all four diets were offered ad libitum to Charolais × Friesian bulls from 187 kg live weight to slaughter at about 488 kg live weight. In experiment 2, the B/PC and B/RSM were offered ad libitum to Charolais × (Hereford × Friesian) bulls from 222 kg live weight to slaughter at about 491 kg live weight. There were no significant differences between treatments in daily live-weight gain, percentage fat and lean in the live animal, slaughter weight or days on experiment in either experiment. In experiment 1, the carcasses from bulls given B/PC were 14 kg heavier than carcasses from bulls given W/RSM (P<0·05), but only 10 and 7 kg heavier than from bulls given B/MG and B/RSM respectively (P<0·05). The killing-out proportion of bulls given B/PC was significantly greater (P<0·05) than bulls given B/RSM. In experiment 2, carcass weights were similar but the killing-out proportion of bulls given B/PC were again greater (P < 0·001). It is concluded that B/MG and B/RSM were as effective as B/PC as finishing diets and that wheat can be successfully substituted for barley, when offered with rapeseed meal.

Hierarchical (each dam mated to one sire) and factorial (each dam mated to several sires) designs involving juvenile selection in a closed nucleus breeding programme were compared for rates of genetic response and inbreeding using stochastic simulation. Numbers of sires and dams selected and herd sizes varied. Generations were discrete. Sires and dams were selected at 15 months on the basis of an estimated breeding value (EBV) calculated using selection index which considered information on close relatives. Selection was for total merit index. Matings were carried out assuming use of multiple emulation and embryo transfer (MOET) or in vitro embryo production (1VEP) technologies, over a range of reproductive parameters. Reproductive rates assumed a fixed number of potential offspring per collection with IVEP and MOET. Comparison of mating designs using MOET showed that rates of genetic response and inbreeding with factorial designs voere from 88 to 431% and 36 to 111% of those with the hierarchical design. Generally rates of genetic response were increased and inbreeding decreased by increasing the number of mates per dam due to a reduction of correlations among EBV. Reduced rates of genetic response were occasionally observed in factorial designs involving the mating of dams to large numbers of sires. This increased the generation interval due to the time to carry out the matings. Factorial designs were found to be most efficient for increasing genetic response and or decreasing inbreeding, for breeding programmes involving the selection of equal numbers of sires and dam. For the breeding programmes considered, rates of genetic response with IVEP equalled the best of the MOET designs, achieved at highest MOET rates, at the lowest level of IVEP. But at the equivalent rate of genetic response, rate of inbreeding was increased by 8%. The increased rate of inbreeding was due to the shorter generation interval with IVEP, and the increased probability of selecting related individuals assuming all donors respond to IVEP.

The possibilities of using embryo transfer (ET) with in-vitro produced embryos as an alternative to artificial insemination (Al) for achieving pregnancy in commercial cattle, and improving the number and quality of calves produced were assessed using 157 multiparous, beefX. dairy suckling cows (mainly Hereford × Friesian) in three autumn and three spring-calving herds in Scotland. Within each herd, cows were allocated at random to (i) Al; or ET with in-vitro produced embryos, either (ii) singly (ET(s)), or (Hi) in pairs (ET(t)) to establish pregnancy. In-vitro produced embryos were derived from oocytes recovered from the ovaries of heifers with high-quality carcasses post slaughter. Semen from a single Simmental sire was used for in-vitro embryo -production and for Al across all herds. Following oestrous synchronization, Al cows were inseminated at 48 and 72 h after removal of the synchrony device. ET cows that had a palpatable corpus luteum received either one or two in-vitro produced embryos placed non-surgically in the ipsilateral uterine horn on day 7 after oestrous (oestrus = day 0). Pregnancy rate, calving rate and calving difficulty were monitored for all cows. Calf performance was monitored on the farms of origin from birth to purchase and thereafter at one finishing unit until slaughter. Animals were slaughtered at a prescribed level of body fatness and saleable meat yields determined.

Cows that received one in-vitro produced embryo gave birth to fewer calves than cows that were artificially inseminated or received two embryos (P < 0·01). Gestation length was longer for cows that gave birth to single ET calves (288 days; P < 0·01) than cows that gave birth to Al calves (285 days) or twin ET calves (284 days). Single ET calves were heaviest at birth (P < 0·01) and higher levels of assistance at calving were required for ET calves whether born as singles or twins. ET calves of both birth types grew more quickly (<1·00 kg/day; P<0·05) and produced larger carcasses (297 kg; P < 0·01) with higher yields of saleable meat (2207 kg; P < 0·01) than Al calves (0·92 kg/day; 281·7 kg; 206·3 kg respectively).

The effect of pregnancy on milk yield was studied on a 601 lactation sample. One hundred and sixty-nine lactations of non-pregnant cows, managed under identical conditions, served as controls. Lactations were divided into five groups according to parity (primiparous or multiparous) and production potential (two ranks for primiparous cows, three ranks for multiparous cows). The difference between the individual milk yield of each pregnant animal, and the mean production of the corresponding controls was computed weekly from the week of conception. The reducing effect of pregnancy began to be measurable from the 20th week of pregnancy, regardless of the group studied or the week of conception. Analysing the amplitude of this effect restricted the investigation to the three following groups: (i) primiparous cows, (ii) low- and medium-producing multiparous cows, (iii) high-producing multiparous cows. The effect of pregnancy was lower in primiparous than in multiparous cows. In the latter, it was higher in high-producing cows, partly because of the greater calf weight. At the 29th pregnancy week, this effect reached -17middot;5, -2·4 and -3·6 kg/day in the three groups, respectively. In each of these three lactation groups, the following non-linear model

Y = −e−a[(Pw~18)e−bpw]

was fitted, where Y is the difference between the milk yields of pregnant and non-pregnant cows at a given stage (week), Pw is the pregnancy week, and a and b are parameters. The overall milk yield losses over a complete lactation, as estimated from the integral of that equation, reached 89, 137 and 203 kg, respectively in the three groups.