Experimental cheeses were prepared in triplicate from pasteurized milk inoculated with Debaryomyces hansenii under aseptic conditions. Three cheesemaking replicates, with efficient control of environmental parameters (temperature, relative humidity, atmospheric composition) showed similar ripening characteristics. Deb. hansenii grew only on the cheese surface, where its oxygen demand was satisfied, especially during the first 24 h (mean generation time, 5·8 h). Salting in a sterile saturated brine solution reduced its growth and decreased viability. Growth was slower after 48 h because of the decrease in ripening temperature (mean generation time, 94 h). The total count of Deb. hansenii was maximum (≈3×107 yeasts/mm2) after 6 d ripening and its viable cell concentration was ≈2×106cfu/mm2. This difference was due to the ‘non-viability’ of part of the population. The viable Deb. hansenii concentration was highly correlated (r2>0·95) with the lactate concentration in the inner part and with the surface and inner lactose concentrations, up to day 10 of ripening. This emphasized the importance of the diffusion of carbon substrate from the inner part to the surface of the cheese during ripening. The pH of the inner part depended significantly on the lactate and lactose concentrations. Surface pH was significantly related to inner lactate concentration, temperature and relative humidity. This also demonstrated the controlling role of carbon source diffusion.

Swiss-type cheeses such as Emmental are characterized by the successive development of thermophilic lactic acid bacteria (TLAB) and propionibacteria. The aim of this study was to determine whether the choice of TLAB strain influenced propionibacteria. TLAB and propionibacteria were cultured sequentially under the conditions prevailing in cheese. Firstly, 11 Emmental juice-like media were prepared by fermenting casein-enriched milk with pure or mixed cultures of TLAB (Lactobacillus helveticus, Lb. delbrueckii subsp. lactis and Streptococcus thermophilus), differing in their proteolytic activities. TLAB cells were then removed by microfiltration. Finally, five strains of Propionibacterium freudenreichii were grown on these media at 24°C under anaerobiosis and their growth characteristics and lactate consumption determined. The media mainly differed in their contents of peptides (1·9–5·3 g/kg) and free amino acids (1·0–5·6 g/kg) and the proportions of lactate isomers (42–92% of the L(+) isomer). Propionibacteria were significantly (P<0·05) influenced by TLAB strains (differences in doubling times of up to 20% and differences in lactate consumption after 600 h culture of up to 52%). The influence of TLAB was similar for all the propionibacteria tested, depended on the TLAB strains and could not be generalized to the TLAB species. Propionibacteria were stimulated by high peptide levels, low levels of free amino acids and NaCl, a low proportion of L(+)-lactate and other undetermined factors. However, variations due to TLAB were less than those between propionibacteria strains.

The concentration ranges in milk and cheese of a number of metallic elements relevant to the health of the consumer are dependent upon animal species, feeding, time of year, environmental conditions and manufacturing processes. We have measured the concentrations of selected trace elements – Al, Ba, Cd, Co, Cr, Cu, Fe, Mg, Mn, Ni, Pb, Pt, Sr and Zn – in raw ewes' milk and typical ewes' milk cheeses. Samples of milk, curd, whey, cheese after moulding, cheese after salting, commercial cheese (Pecorino), Ricotta, scotta and brine were analysed for seven complete cycles of cheese production. Samples of curd, whey, brine and waste product (i.e. scotta) were collected and analysed in addition to the cheese and milk to assess the distribution of each element in the main product, by-products and waste products during cheesemaking. Quantitative determinations were performed mainly by inductively coupled plasma atomic emission spectrometry. There were considerable differences among the trace element contents of raw ewes' milk and related products, indicating that manufacturing processes play a key role in the distribution of minor and trace elements. Two different approaches were used to assess the dietary intake of these elements associated with the consumption of Pecorino and Ricotta, and whether this had any nutritional or toxicological implications. The results indicated no toxicological risk and showed that, on the contrary, Pecorino may be a good source of several essential elements.

Mammary infections caused by Staphylococcus aureus are still one of the most serious problems in dairy farms all over the world (Sischo et al. 1993), and the epidemiology of the infection has not yet been completely elucidated (Aarestrup et al. 1995). Any effective modern approach to this disease must therefore be based on more comprehensive epidemiological studies, conducted with valid microbiological typing tools.

A technique for use in epidemiological studies should identify many types, and should be inexpensive, quick and easy to perform, but above all reproducible. Among the available methods, phage typing has up to now been widely and successfully used in differentiating strains of Staph. aureus isolated from cattle with mastitis (Mackie et al. 1987; Fox et al. 1991), but it has some limitations, being a technically demanding method subject to considerable experimental as well as biological variation (Maslow et al. 1993). Moreover, in some studies the number of strains that could not be typed with available bacteriophage panels has been high (Carroll & Francis, 1985; Farah et al. 1988).

Alternative methods have been investigated, and of these molecular techniques have been the most intensely studied (Aarestrup et al. 1995). In studies of human infections caused by Staph. aureus, analysis of the so-called X region of protein A gene polymorphism has been a useful epidemiological marker (Frénay et al. 1994). This gene is ∼2150 bp and harbours some functionally distinct regions: an FC-binding region, the so-called X region and, at the C terminus, a sequence required for cell wall attachment (Guss et al. 1984; Frénay et al. 1994). The X region polymorphism depends on the presence, within the region itself, of a varying number of 24 bp repeats, highlighted by the amplification of this highly polymorphic DNA region and its subsequent restriction fragment length polymorphism (RFLP) analysis (Frénay et al. 1994).

Human epidemic (MRSA) and non-epidemic methicillin-resistant Staph. aureus (non-MRSA) strains, which both cause infections but have completely different infection patterns, have been successfully distinguished by analysis of this polymorphism (Frénay et al. 1994). However, protein A has been identified in only 93% of Staph. aureus strains isolated from bovine intramammary infections (Poutrel & Ducelliez, 1979; Johne & Jarp, 1988).

The aim of the present study was to determine whether the gene for protein A of Staph. aureus (Spa) was present in Staph. aureus strains isolated from cases of subclinical bovine mastitis. This was carried out using the polymerase chain reaction (PCR), as suggested by Frénay et al. (1994). In addition, we have investigated the genetic polymorphism related to the X region of the gene, by means of PCR amplification and subsequent RFLP analysis. Finally we verified the stability of this region after in vitro subculture.

Branched-chain amino acids (AA) are mostly metabolized in the splanchnic area, but some are metabolized within the mammary gland and thus could contribute to the synthesis of non-essential AA (Wohlt et al. 1977). The extraction rate of leucine from plasma by the mammary gland is particularly high (64% in the goat; Roets et al. 1983), in excess of that used for the synthesis of milk proteins (Davis & Mepham, 1976; Wohlt et al. 1977; Roets et al. 1983). Thus, although mammary leucine is mainly used for milk protein, it also provides amino nitrogen and carbon sources for the synthesis of non-essential AA.

To our knowledge, no information is available on the transfer and distribution of plasma leucine amino nitrogen to milk protein AA. Using the technique for chromatographic fractionation of AA recently developed by Casseron et al. (1997), we studied the specific 15N enrichment of casein (CN) AA in the goat given a single intravenous injection of [15N]leucine.

A number of endemic diseases of dairy cattle cause significant losses to the dairy industry in the mainland UK (England, Scotland and Wales), both in terms of the reductions in output levels or wastage of resources incurred and the resource costs of disease prevention and treatment (Esslemont & Spincer, 1993; Esslemont & Kossaibati, 1996). Various studies have estimated the costs associated with different diseases (Bennett, 1992). However, these studies use different methods of assessment, relate to different populations at risk, refer to different points in time and utilize different ways of measuring disease and valuations of the effects of disease on production. Thus, it is difficult to use these studies for any comparative assessment of the magnitude of output losses and resource wastage incurred as a result of different diseases. Such information is useful in exploring both the economic consequences of diseases and the potential benefits of research on improved disease control (Howe, 1991; McInerney, 1996).

This paper presents analyses of the impacts on production of five endemic diseases and conditions of dairy cattle in mainland UK: bovine viral diarrhoea (BVD), fasciolosis, lameness, leptospirosis and mastitis (including summer mastitis). These analyses follow from a preliminary economic study of the impacts on livestock production of some 30 non-notifiable diseases and conditions of farm animals (Bennett et al. 1997). The study was funded by the Ministry of Agriculture, Fisheries and Food in the UK, with the (eventual) aim of providing information to policy makers that might help them to reach decisions on allocating funds to research into livestock diseases. Full details of the analyses are available from the website address given at the end of this paper.

The light stabilities of two commercial colorants for Cheddar cheese, an annatto extract and a β-carotene preparation, have been compared in a model system and tested in industrial scale production. The apparent quantum yields for photodegradation by monochromatic light were determined at 25 °C for solutions of the colorants in air-saturated 10 mm-citric acid–disodium hydrogen phosphate buffer containing 100 mg sodium caseinate/l at pH 5·2 and 5·4. Solutions of commercial β-carotene were more photostable than solutions of annatto, and although the quantum yields for photodegradation for both solutions depended significantly on both pH and irradiation wavelength, exposure to u.v. light (313 and 366 nm) caused more photobleaching than exposure to visible light (436 nm). The differences in colour stability observed in the model system were verified by tristimulus colorimetry during 9 weeks illuminated storage at 5 °C of Cheddar cheese produced from milk to which had been added (per 1000 kg) 67 g β-carotene extract or 133 ml annatto.

The reduced sulphur volatiles, methanethiol and dimethyldisulphide (but not dimethylsulphide), have been detected in the headspace of butter samples. Their concentrations in the butter itself were calculated by determination of the distribution coefficient (K) through spiking butter samples with known quantities of methanethiol and dimethyldisulphide. The K values obtained suggest that the dimethyldisulphide had a greater affinity than methanethiol for the lipid phase of the butter. Solid phase microextraction used in conjunction with gas chromatography–mass spectrometry has been shown to be an effective method for headspace analysis and was used for quantifying the butter headspace concentrations of the reduced sulphur compounds. Seasonal variation of both methanethiol and dimethyldisulphide were found in a series of butter samples, the highest concentration being in spring with an overall decrease through the New Zealand summer. Grass type and condition appeared to influence the concentrations of these compounds in the butter. Storage of the series of butter samples at 4°C for 5 weeks resulted in a significant decrease in the concentrations of both methanethiol and dimethyldisulphide. The dimethylsulphide concentration in all butter samples was below the detection limit of the analytical system used.

An automatic milking system (AMS) uses a robot to attach the teatcups and can be continuously available for cows to present themselves to be milked with minimal human intervention. Milking in an AMS is therefore a continuous rather than a batch process, and the conventional practice of cleaning a milking system after each batch of cows has been milked is not applicable.

Milk flow through an AMS is different in scale and in temporal distribution from that in a conventional system. These differences in flow patterns may be significant for hygiene. It has been suggested by Verheij (1992), for example, that milk will flush away much of the residue of the previous milking and that microbial growth will be retarded because of the bacteriostatic properties of fresh milk. Ordolff & Bölling (1992) reported experiments in which batches of milk were passed through a milking installation that was said to simulate an AMS. Their main conclusion was that milk residues left in the system for up to 60 min did not increase the bacterial contamination of the next batch of milk. However, the practical relevance of this conclusion is not clear because the design of the simulated AMS was not specified, so it can not be compared to a practical AMS. Moreover, mixtures of fresh and UHT milk were used, with unknown consequences. The purpose of the experiment reported in this paper was to strengthen the evidence on which to base recommendations for effective cleaning procedures for AMS.

The bacterial count of milk that had been passed through a system was used as the indicator of the cleanliness of the system. It was apparent from previous work by Ordolff & Bölling (1992) that this would produce results that varied widely. The experiment was therefore designed to be as simple as possible. In particular, a simple, easily defined laboratory rig was used rather than an AMS.

Lipid oxidation in milk and dairy products is a chain reaction initiated by formation of free radicals (Richardson & Korycka-Dahl, 1983). Thanks to intensive studies on both model systems and actual food, the autocatalytic process, including the formation of secondary lipid oxidation products from the lipid hydroperoxides formed initially, is fairly well understood. However, actually predicting the rate at which the first free radicals leading to spontaneous oxidation are formed in milk from different cows awaits the development of new analytical methods with higher specificity and sensitivity (Nicholson, 1993; Barrefors et al. 1995). Such methods would also be valuable for predicting the stability and shelf life of dried dairy products, which are determined by oxidative phenomena. Electron spin resonance (ESR) spectrometry has the potential for detecting the early events in lipid oxidation, as it is the only spectrometric method that will directly detect the unpaired electron characteristic of the free radical and it is, moreover, a highly sensitive method (Brudvig, 1995). ESR spectrometry has recently been shown to provide quantitative information on the level of free radicals in milk powder that correlates with the level of secondary oxidation products developed upon reconstitution and that also correlates with subsequent sensory evaluation (Nielsen et al. 1997; Stapelfeldt et al. 1997a, b, c). However, in order to explore further the potential of this method for raw milk, it was considered valuable to measure the tendency of milk to form free radicals in relation to its level of α-tocopherol, the most important lipophilic chain-breaking antioxidant (cf. Kamal-Eldin & Appelqvist, 1996).

The use of transgenic animals to manipulate milk composition has considerable potential, both for the production of biomedical proteins and for the direct manipulation of milk composition for the improvement of dairy animals and their products (for reviews, see Wall et al. 1992; Yom & Bremel, 1993). Promoters from a number of milk protein genes from a variety of species have been tested for their ability to direct the expression of foreign proteins to the mammary gland (for review, see Maga & Murray, 1995).

β-Lactoglobulin (β-lg) is the major whey protein produced in ruminant milk and is part of the normal milk composition of most mammals except humans and rodents (Pervaiz & Brew, 1985). It is expressed at high levels in the mammary gland and is developmentally regulated. Transgenic mice have been produced using the complete ovine (Simons et al. 1987; Shani et al. 1992) and caprine (Ibañez et al. 1997) β-lg genes. In general, high levels of expression were obtained with the ovine β-lg gene, and expression was also seen in a position-independent manner (Whitelaw et al. 1992). Lower levels of expression were reported using the caprine β-lg gene. Here we report the production of transgenic mice using the bovine β-lg gene. We describe high expression, position-dependent, and copy number-related expression of bovine β-lg protein in the milk of six lines of transgenic mice.

The sensory and rheological characteristics of stirred yogurts varying in fermentation temperature, heat treatment of milk, dry matter content and composition of bacterial cultures were investigated. Two independent sensory properties, non-oral viscosity and oral viscosity, were evaluated. To reflect these properties precisely it was necessary to perform rheological measurements at both small and large deformations. Non-oral and oral viscosities were successfully modelled from a set of rheological variables by partial least squares regression, resulting in a three component model explaining 83·8% of the variation in non-oral viscosity and a two component model explaining 82·0% of the variation in oral viscosity. Non-oral viscosity was highly correlated with the complex modulus (G*) from the dynamic oscillatory measurements (r=0·823) and the viscosity obtained from a Brookfield viscometer (with Helipath stand) operating at 5 rev./min (r=0·862). Similarly, there were relationships between oral viscosity and the hysteresis loop area from the viscometry test (r=0·867) and between oral viscosity and the viscosity obtained from the Posthumus funnel (r=0·834).

In the lactating mammary gland, epithelial cells secrete triacylglycerols in the form of droplets enveloped by an apical surface membrane. This membrane is known as the milk fat globule membrane (MFGM; Mather & Keenan, 1983). MFGM-associated proteins have been studied and employed in developing antibodies against surface antigens of breast epithelial cells, which are used in breast cancer immunodiagnosis and histopathology (Salinas et al. 1987; Larocca et al. 1991; Peterson et al. 1995). So far, only a small number of proteins have been analysed and their sequence identified in bovine, murine and human MFGM; among them are butyrophilin (Jack & Mather, 1990; Ishii et al. 1995; Taylor et al. 1996), MFG-E8 (Stubbs et al. 1990), PAS 6/7 (Hvarregaard et al. 1996) and lactadherin or breast antigen BA46 (Couto et al. 1996; Taylor et al. 1997). Several minor proteins have yet to be characterized, since it is not easy to isolate them in large quantities from the membrane. SDS gel patterns give useful information about MFGM proteins, such as concentration, relative molecular mass and presence of carbohydrate. Over forty membrane components have been separated by electrophoretic techniques from bovine MFGM (Mather et al. 1980).

The research reported here combined SDS-PAGE with sequencing analysis and describes the composition of human MFGM, with the exception of high molecular mass mucin, which only penetrates an acrylamide gel of 40 g/l. Mucins have been extensively studied and the sequence predicted from cDNA (Gendler et al. 1990). Surprisingly, identification of the protein bands in the present study revealed that three proteins alone constituted the major components of human MFGM: xanthine oxidase (EC 1.1.3.22), butyrophilin and lactadherin. Lactadherin belongs to a family of proteins possessing epidermal growth factor-like and factor V/VIII C1/C2-like domains, including bovine PAS 6/7, guinea pig GP55 (Hvarregaard et al. 1996) and murine MFG-E8 (Stubbs et al. 1990). In a previous investigation, we characterized lactadherin (formerly breast antigen BA46) and its truncated 30 kDa form as components of healthy human MFGM (Giuffrida et al. 1998). Human butyrophilin has recently been cloned and sequenced (Taylor et al. 1996); the presence of two extracellular immunoglobulin superfamily domains suggested a potential cell surface receptor function. This study was aimed at identifying and characterizing the multiple forms of the major proteins of MFGM.

The casein micelles of bovine milk are large aggregates of roughly spherical shape consisting of the four phosphoproteins αs1-, αs2-, β- and κ-casein (CN). They are polydisperse colloidal particles with weight-average diameters between about 50 and 500 nm (Horne, 1984). Since αs1-, αs2- and β-CN are sensitive to calcium it is assumed that they are located predominantly within the micelle, whereas κ-CN is mainly on the surface of the micelles and serves as a stabilizer for the aggregates owing to its hydrophilic C-terminus. Inorganic calcium phosphate interacts with the phosphate groups of the caseins. This conformation of caseins, calcium and phosphate also contains small amounts of magnesium and citrate. Of the dry matter of the micelle, ∼93% is casein and ∼7% inorganic components, mainly calcium and phosphate (McMahon & Brown, 1984; Holt & Horne, 1996).

In several studies it has been demonstrated that milk with the κ-CN B variant has on average smaller micelles than κ-CN A milk (Morini et al. 1975; Nuyts Petit et al. 1997). Compared with κ-CN variants, there are only a few studies on the association of αs1-, αs2- and β-CN variants with casein content and micelle size. In the milk of β-casein-deficient mice, casein micelles were smaller than in normal mouse milk but still present. The lack of β-CN was partly compensated by an increased whey protein content (Kumar et al. 1994). A comparison of the cheesemaking properties of Tarentaise milks containing the A (A1 or A2, without distinguishing between these variants) and C variants of β-CN revealed that the most significant characteristic for discriminating the two types of milk is the average diameter of the casein micelles: 293 nm in β-CN C type milk, 187 nm in the A type (Delacroix-Buchet & Marie, 1994).

The objective of the present study was to investigate the micelle size and the casein composition of milk from half sibs within Simmentals cross bred with Red Holsteins (Simmentaler Fleckvieh, SI×RH) sires heterozygous for A1A2 of β-CN.

Rotavirus is the leading cause of infectious diarrhoea in infants and children throughout the world (Davidson, 1996). In developed countries, rotavirus is associated with significant morbidity, while it causes a high rate of infant mortality in developing countries (Blacklow & Greenberg, 1991).

At present there is no effective treatment or vaccine against human rotavirus. Although considerable effort is being put into developing such a vaccine (Molyneaux, 1995; Midthun & Kapikian, 1996), researchers are encountering difficulty in obtaining sufficiently broad protection against the great variety of human rotavirus serotypes (Kapikian & Chanock, 1985). Passive immunization has been tried as an alternative, and clinical trials have shown that oral intake of bovine colostrum containing anti-human rotavirus antibodies is effective in preventing rotavirus infection and transmission in experimental animals, infants and children (Davidson, 1996). Since cross reactivity has been found between bovine and human rotavirus strains (Kapikian & Chanock, 1985), a product containing specific antibodies against human or bovine rotavirus strains may exert a protective effect in human babies. Such a product would be a useful temporary measure while an effective vaccine is developed. Furthermore, these products could be valuable for immunocompromised patients in preventing rotavirus infections.

However, such products would be subjected to processes such as pasteurization or ultra high temperature treatments, so it would be necessary to know the effect of heat on the biological activity of the antibodies. We have previously studied the effect of heat on the structure of bovine colostrum immunoglobulins IgG, IgA and IgM and evaluated the kinetics of heat denaturation by measuring the loss of their ability to bind to specific antibodies directed against them (Mainer et al. 1997). In the present study we have investigated the effect of different heat treatments on the biological activity of immunoglobulins from bovine colostrum by measuring the virus-neutralizing activity (VNA) of heat-treated colostrum containing anti-rotavirus immunoglobulins against a bovine strain. In addition, we have studied the distribution of the VNA between the different classes of immunoglobulins of the anti-rotavirus colostrum.

We have reported previously on the kinetics of thermal inactivation at 80–120°C of the extracellular proteinase from Pseudomonas fluorescens 22F (Schokker & van Boekel, 1997, 1999b). During these studies, we noted some inactivation during the heating up and cooling periods, but allowed for this by calculating the residual activity as a fraction of the activity after the heating up period of 2 min followed by cooling to 0°C. However, it may be of interest to evaluate the extent of inactivation during these heating up and cooling periods. If the temperature dependence of the reaction rate behaves according to Eyring's theory, inactivation would, of course, be slower than at the final heating temperature. However, during the heating and cooling of the enzyme solution, the temperature also passes the region in which autoproteolysis occurs (Schokker & van Boekel, 1998a). Prolonged residence time in the critical zone for autoproteolysis may cause increased inactivation, as has been demonstrated in electrophoresis experiments for proteinases from other Ps. fluorescens strains (Barach & Adams, 1977; Richardson, 1981; Diermayr et al. 1987). Consequently, the inactivation during the first few minutes would be dependent on factors influencing both autoproteolytic and thermal inactivation.

In most of our heating experiments (Schokker & van Boekel, 1997, 1999b), inactivation during heating up was relatively rapid compared with inactivation at the final heating temperature, leading to a biphasic inactivation curve. This was also found for proteinases from many other Ps. fluorescens strains. In some studies the inactivation during heating up was not taken into account when analysing the kinetics of thermal inactivation (Patel et al. 1983; Yan et al. 1985; Fairbairn & Law, 1986), which led to misinterpretation of the mechanism or the kinetic values. Others explained the biphasic inactivation curve by autoproteolysis (Barach & Adams, 1977; Richardson, 1981; Stepaniak & Fox, 1983; Kroll & Klostermeyer, 1984; Diermayr et al. 1987), or stabilization by Ca2+ of a small portion of the proteinase to heat inactivation (Stepaniak & Fox, 1983; Azcona et al. 1988).

In this paper we discuss the influence of protein, enzyme purification and Ca2+ activity on inactivation during the heating up and cooling periods. The aim of this study was to determine, using kinetic modelling, whether the inactivation during heating up and cooling periods could be explained by autoproteolysis and thermal inactivation, or whether other mechanisms are involved in the strong initial inactivation.

Ion exchange is used commercially in such processes as water softening, chemical purification, separation of ionic from non-ionic forms and analytical applications. Van Kreveld & van Minnen (1955) reported a method for the determination of Ca2+ and Mg2+ activity in milk using an ion-exchange resin. From this early work, it was established that with suitable resins exchange of salts was possible in milk, permeate and whey. This included the exchange of Ca and Mg from micelles by using Na+ or K+. Cationic resins of the Zerolit 236 type have been used successfully in the production of Ca-reduced single cream (Anderson et al. 1976). Such treatment prior to UHT processing was shown to increase physical stability when it was used as a coffee whitener. There is evidence that reduction of Ca2+ would also reduce the deposit formation that may occur during storage of UHT milk. Thus there is a need to establish commercial techniques to produce Ca-adjusted milks that could then be used as raw materials for industrial applications or product development work.

The present study investigated the use of ion exchange to reduce Ca in milk with particular reference to understanding the exchange of salts in Ca-reduced milk and the changes to some of its physical properties.

Dr M. Elisabeth Sharpe, who died last year, was Editor of the Journal of Dairy Research from 1975 to 1989. During this period she devoted her considerable energies and talent to putting the Journal on a sound financial footing and expanding its contributions from authors in the developing world. Becky, as she was known to her friends and colleagues, was highly successful in both these goals and the Journal continues to build upon her success to this day.

Her editorship was the culmination of a notable scientific career, full of achievement and friendships. Born in 1916 in Barnsley, Yorkshire, Becky Sharpe completed her BSc degree at University College London in 1937 and went on to a year of postgraduate study in microbiology at the University of Manchester. She then joined the staff of the National Institute for Research in Dairying, Shinfield, an Agricultural Research Council Institute and a part of the University of Reading. She left in 1942 for a 4 year stint with The Boots Company in Nottingham and returned to Shinfield for a 30 year period of exemplary research in dairy microbiology in the Department of Bacteriology.

The Department had a tremendous group of researchers and was a really exciting place for many of us to start our careers working with people like Becky Sharpe, Bruno Reiter, Frank Neave and Christina Cousins. Becky interacted with many of us over her career, and her research resulted in hundreds of publications and contributions to the scientific literature on the microbiology and microflora of milk and dairy products. She is probably best remembered for her work on staphylococci, micrococci and the lactobacilli. Often she pioneered the development of new techniques for the growth, isolation and identification of these Gram-positive organisms. She was especially pleased with her work as a contributor and editor for Bergey's Manual of Systematic Bacteriology, which continued after retirement. She received her PhD from the University of Reading in 1951 and a DSc degree from the University of London in 1973.

On a more personal note, the author remembers Becky as a generous and stimulating scientific colleague. She was always willing to listen to the ideas of a young and inexperienced scientist and offer help and advice. She was highly respected and well known in the world of dairy microbiology, and many outstanding microbiologists either worked with her or trained under her. Her laboratory was a favoured stop for researchers from across the globe.

After retiring from active research in 1976 she remained very active in scientific circles. Her Editorship of the Journal was a major commitment, but she was also an active participant in scientific society activities, particularly as a fellow of the Institute of Biology. Her numerous scientific contributions, and the health and international flavour of the Journal speak to her professional accomplishments. Her friendship and help will be fondly remembered and missed by all of those fortunate enough to have worked with her.

Enumeration of bacteria in raw milk is of public health and economic importance. Among the proposed rapid methods for assessment of bacterial contamination in raw milk, ATP bioluminescence has proved to be one of the most promising (Griffiths, 1991). Several companies produce ATP bioluminescence reagent kits and equipment for analysing raw milk samples for total bacterial count (Sutherland et al. 1994; Reybroeck & Schram, 1995). The principle of ATP bioluminescent bacterial assay is based on the following assumptions (Olsen, 1991). All living organisms contain ATP, ATP is neither associated with dead cells nor absorbed on to surfaces, colloids and so on, and there is a fairly constant ratio of ATP to biomass/number of cells for all microbial taxa independent of metabolic activity or environmental conditions. Of these assumptions, only the first seems to be indisputable. It is not the number of bacterial cells, but rather the colony forming unit (cfu) that is the denomination used when assessing the microbial quality of milk. For Gram-negative rods of the genus Enterobacteriaceae, a cfu is usually derived from a single cell. However, Gram-positive cocci (staphylococci and streptococci) grow in bunches and chains respectively (Gregg, 1991), and estimation of cell numbers may not give good agreement with the colony counts.

Several approaches have been investigated to increase the sensitivity of the bioluminescent method (Pahuski et al. 1991; Sutherland et al. 1994; Reybroeck & Schram, 1995; Froundjian et al. 1999). Although the detection limit achieved by these modifications (104 cfu/ml) may be sufficient for practical use (Bautista et al. 1992; Reybroeck & Schram 1995), the accuracy of the analysis was not significantly improved. The reported values for accuracy of the estimate for cfu/ml in raw milk (Syx) by the bioluminescent method were in the range 0·27–0·87 log units (Bautista et al. 1992; Reybroeck & Schram, 1995). The purpose of the present study was to determine the reasons for the lack of accuracy of the bacterial ATP assay in raw milk.

Lactobacilli are of great commercial value because they are widely used as starters in food fermentation. The species Lactobacillus delbrueckii, which comprises the three subspecies bulgaricus, lactis and delbrueckii, is important in dairy products and vegetables. The subspecies bulgaricus is present in yogurt, and subspecies lactis is recovered from whey, starter cultures and cheeses (Stiles & Holzapfel, 1997). It is unusual to recover Lb. delbrueckii subsp. delbrueckii (Lb. delbrueckii) from dairy products, from which Lb. delbrueckii subsp. lactis (Lb. lactis) and subsp. bulgaricus (Lb. bulgaricus) are typically isolated. However, given the similarity of these two latter subspecies, a clear identification of isolates on the basis of phenotype criteria alone is often problematic (Dellaglio, 1989; Millière et al. 1996).

The use of DNA probes and methods based on polymerase chain reaction (PCR) have greatly facilitated diagnostic identification of bacteria. For the lactobacilli, DNA probes have been described for Lb. helveticus, Lb. acidophilus, Lb. fermentum, Lb. plantarum and most of the other Lactobacillus species (Pot et al. 1994; Tailliez et al. 1994; Quere et al. 1997). For Lb. delbrueckii, an EcoRI DNA fragment of the plasmid pY85 was used as a probe for this species, although it was not able to discriminate its three subspecies (Delley et al. 1990).

In our laboratory, a specific amplification of a DNA fragment of ∼1·7 kbp using universal primers for the amplification of ribosomal RNA (rRNA) genes was found only for dairy isolates of Lb. lactis and not for Lb. bulgaricus, Lb. delbrueckii, Lb. helveticus and Lb. acidophilus (G. Giraffa, P. de Vecchi and L. Rossetti, unpublished results). This prompted us to test the possible use of this fragment as a specific DNA probe for Lb. lactis. To this end, Southern and dot blot hybridization experiments were carried out with total DNA of several strains belonging to different lactic acid bacteria (LAB) species.