A description is given of various methods for the ultrafiltration and dialysis of milk and of the composition of the sera obtained. Ultrafiltrate prepared by the procedure recommended is reasonably representative of the aqueous phase of milk, but its content of lactose and citric acid, and consequently also of calcium, is determined to a slight degree by the sieving phenomenon known to occur often in ultrafiltration. The composition of diffusate obtained from milk at 20 °C is not thought to be controlled to any significant extent by a Donnan effect and is regarded as identical with that of the aqueous phase of milk. The lactose content of diffusate suggests that about 2 % of the water in milk is bound to protein, and allowance should be made for this when calculating the concentrations of the soluble constituents in milk from the composition of diffusate. Diffusate prepared from milk at 3°C contains slightly more total calcium, ionized calcium and phosphorus than diffusate prepared at 20 °C. These differences are attributed to a change in the partition of calcium and phosphorus between the disperse and aqueous phases at the lower temperature, an explanation that is supported by the reversibility of the change. The composition of diffusate prepared by the procedure recommended indicates that about 5 % of the sodium and about 6 % of the potassium and citric acid in milk are in the. disperse phase.It is sometimes desirable to know not only the total concentration in milk of a given constituent but also the amounts of it in the disperse and aqueous phases, i.e. how much can be regarded as ' colloidal' and how much as ' soluble'. This information is usually obtained by determining the total amount of the constituent and the amount in solution; the colloidal fraction is then obtained by difference. Such a fractionation depends on the preparation of a suitable serum for estimating the soluble constituents. This serum should have the same composition as the aqueous phase of milk and therefore in its preparation the equilibria in milk should be disturbed as little as possible. Several methods of preparing a serum are available and of these the most suitable would appear to be ultrafiltration or dialysis. Other methods of estimating the soluble constituents of milk are the analysis of rennet whey or the serum obtained by high-speed centrifuging, and the treatment of milk with ionexchange resins.

SummaryA subjective test for the determination of the stability of milk protein to heat is described. In the test, the time required for particles of coagulated protein to become visible throughout a 2·5-ml sample of separated milk maintained at 135°C in a glass tube rocking at 8 c/min is taken as a measure of stability. The precision of the test was such that single determinations were generally adequate.Coagulation time decreased by about 12% as rocking speed was increased over the range 4–12 c/min and increased by a factor of about 3 for a decrease in heating temperature of 10 degC over the range 140–105 °C; with some milks the Q10 °C value increased to 5–8 a temperature decreased. As sample volume was increased over the range 1–3 ml coagulation time increased, especially With milks whose coagulation was poor (initial clots small). This volume effect appeared to be a consequence of the accompanying decrease in the proportion of headspace oxygen to volume of milk.

1. This paper is Part I of a series dealing with the relation between the chemical composition of milk and the stability of the caseinate complex to ethanol (Part II), rennet (Part III) and heat (Part IV).2. A description is given of the milk samples, which were taken from herd bulk milk, from individual cows in different stages of lactation and also from cows with subclinical mastitis.3. The methods used to measure the stability of the caseinate complex to the three coagulating agents are described, together with the methods used to make a detailed chemical analysis of the milk samples.4. The chemical composition of the samples is given and also the relation between composition and stage of lactation, and the interrelations of the concentrations of certain milk constituents.The authors thank Miss M. H. King and Miss R. E. M. Stevenson for technical assistance, Mr N. H. Strachan and Mr R. C. Voss (West of Scotland Agricultural College) for the sodium and potassium determinations and Dr P. S. Blackburn for the differential cell counts and bacteriological examination of the milks.

Coagulation time-pH curves with a coagulation-time minimum around pH 6-8 (type A curve) could progressively become type B (no minimum) as the heating temperature was decreased from 150 to 130 °C. The short coagulation time that most milks have when pH is around 6-8 was found to be the result of a 'premature' coagulation, probably caused by calcium phosphate deposition on the larger caseinate micelles. This is followed by a second coagulation, not visually detected, that coincides with the coagulation time that would be expected if no coagulation-time minimum existed on the coagulation time-pH curve. The coagulation time of milks giving type A and type B curves may therefore not be comparable.Forewarming milk for 30 min at 80 °C can introduce or accentuate a coagulationtime minimum when the milk is subsequently heated at a higher temperature. The effects of adding /?-lactoglobulin, copper and iV-ethylmaleimide on the heat stability of milk were examined and explanations proposed for these effects.In their studies on the stability of milk protein to heat, Davies & White (1966a, b) and White & Davies (1966) found that the coagulation time of milk as determined by their subjective heat-stability test could vary with the proportion of headspace O 2 to volume of milk and the age of the milk. Their objective heat-stability test revealed that 'normal' or 'abnormal' protein coagulation could occur in different milks, the latter apparently being a 2-stage process and often found with milk from cows with sub-clinical mastitis. The object of the present series of investigations was to elucidate further factors thought to govern the heat stability of milk, paying special attention to the well-known high sensitivity of coagulation time to pH (Rose, 1961a, b) EXPERIMENTAL Samples of whole milk were obtained from the Hannah Institute herd of Ayrshire cows, usually from individual cows but sometimes from the herd bulk milk, and

1. During 1950–53 the milk of 814 Ayrshire cows was sampled six times during the lactation at intervals of approximately 5 weeks, starting towards the end of the first month of lactation. The weight of milk at successive evening and morning milkings was recorded and samples from each mixed in proportion to the yield. The milk was analysed for total solids, fat, S.N.F. by difference, crude protein (N×6·38), casein and lactose.2. The weighted lactation average was calculated for each constituent of the milk of each cow and used in genetic studies (see Part II), and in assessing the effect of age on milk composition. The analyses of the individual samples were used to determine the separate effects of stage of lactation and of season on milk composition.3. Advancing lactation caused the following changes in milk composition:(a) Milk yield was highest 45 days after calving and then fell regularly to the end of lactation.(b) Total solids, S.N.F., fat, crude protein and casein contents fell rapidly for 45 days, with fat and total solids continuing to fall for a further 30 days. The concentrations of all these constituents then increased continuously for the remainder of the lactation, rising more rapidly after about 200 days.(c) The changes in lactose content were opposite from those of fat and protein and smaller in magnitude. The changes bore a marked resemblance to those for yield. The value rose to a maximum after 45 days, fell slowly until about 165 days after calving and then more quickly. The lactose content of milk from cows in their first lactation fell much more slowly with advancing lactation than that in the milk of older cows.4. Seasonal effects were of smaller magnitude than those arising from advancing lactation and caused the following changes in milk composition:(a) Yield rose steadily from January to the May-June period, and then fell to a minimum during October-November.(b) Fat content was at a maximum in October, falling steadily to a minimum in June.(c) Crude protein and casein contents rose to a peak in May-June and again in September. Lowest values occurred from January to March.(d) Lactose content was at a steady high level from January to June, falling to a lower level by August at which it remained for the rest of the year. Again there was some similarity in the pattern of change in lactose content and in yield. The range of variation in the values for lactose was less than those for protein and fat, and the variations were mainly opposite in sign.(e) Total solids and S.N.F. were at a minimum in March and April, but whereas S.N.F. reached their highest level in May-June, total solids were highest in October.5. Increasing age of the cow resulted generally in increasing yields of poorer quality milk. The difference in composition between milk from the first and the grouped ninth and later lactations was fat 0·19%, S.N.F. 0·34%, crude protein 0·08%, casein 0·21% and lactose 0·25%.6. On the evidence of cell counts, a number of the samples came from cows with some inflammation of the udder. The average casein numbers, however, did not indicate any serious incidence of mastitis, and it was concluded that although lactose contents may have been lowered slightly, the effect of disease on the average composition was small.7. Although the figures are not closely comparable, it appears that there has been little change in the composition of the milk of Ayrshire cows in Scotland since 1921–22.8. The number of individual cow samples deficient in fat was 2% and in S.N.F. 16% of the total. In 560 samples of farm bulk milk none were deficient in fat and only 1·6% deficient in S.N.F. May, June and July were the months with most fat-deficient samples, and the period from January to April provided the largest number of samples deficient in S.N.F., closely followed by the period July to October.9. Possible causes of the seasonal changes in milk composition are discussed. These changes are thought to be caused mainly by feeding although this does not give a complete explanation of all variations.