Use and users of artificial insemination in Swedish dairy cattle breeding, 1935–1955

ABSTRACT From the middle of the twentieth century, the new technologies and techniques of artificial insemination (a.i.) transformed dairy cattle husbandry and breeding across dairy-producing countries. While there are nuanced and multi-faceted studies of early a.i., previous work has not engaged much with its material usage, meaning that we know little about how different techniques, practices, and animal reactions promoted or restricted a.i’.s use for particular purposes. Here, I address this aspect by studying early a.i. in Swedish cattle breeding as a concrete set of technical artefacts used by humans and animals, thus reframing a.i’.s early history as a problem of use and users. I focus specifically on the artificial vagina, the predominant instrument used to gather bull semen, and show how the modes of use and non-use of the artificial vagina not only helped shape a.i. itself but also veterinary expertise, the institutions of breeding, and parts of the Swedish agrarian economy.


Introduction
From the middle of the twentieth century, the new technologies and techniques of artificial insemination (a.i.) transformed dairy cattle husbandry and breeding across dairy-producing countries. Enabling humans to insert pre-collected semen into cows' reproductive tracts, a.i. disaggregated fertilization from the act of mating. The impregnations necessary for milk production could henceforth be organized in a more rational manner than before. Furthermore, a.i. opened up new ways to identify and intensively use bulls carrying good traits for milk and butterfat yields. As historian Deborah Valenze has observed, it therefore became a key technology in the development of today's highoutput dairy economies of scale. 1 Here, I address the development and introduction of this formative agricultural technology in Sweden from a hitherto-unexamined point of view, examining how both human and bovine users related and reacted to artificial insemination technologies and what effects their forms of usage had on a.i'.s development and integration into farming. Earlier work on a.i. in dairying in Sweden and elsewhere, of which I would particularly highlight agronomist Jan Rendel's and historian of science Sarah Wilmot's solid and multi-faceted accounts, generally write the technology into broader histories of several of the involved institutions 10 -I show how particular modes of use -human and animal -of a.i. technology shaped a.i'.s development, organization and impact.
In science and technology studies, the relevance of considering the active role of users in technological development has been established at least two decades. STS scholars in this field reject essentialist understandings of use and users as being inherent in technological artefacts. They seek to understand how various artefacts are used in practice, and how designers' ideas about use -materially inscribed in their designs -are either taken up or challenged by users, highlighting the latter's role in shaping and co-constructing technologies as they develop. 11 STS research tends, however, to assume a human user, whereas I want to consider animal use as well. This also puts the paper in relation to a growing literature discussing human-animal relations in history. 12 I do not draw extensively on this body of work, but share some of its starting points in that I believe that it is necessary to look at a.i. bulls as acting subjects who behaved according to their interests, and to consider how their actions (or inactions) produced historically significant effects. 13 Attending to human-animal interaction in early a.i. enables me to make several novel points about these effects. First, I show that when a.i. was introduced in Swedish farming from 1943, a large number of bulls either rejected the artificial vagina outright or used it in a manner that prevented semen of sufficient quality from being gathered. The organization of early a.i. in Sweden, and the expertise that gathered around it, was defined largely by these serious and pervasive failures. Second, this leads to an insight of general applicability to a.i. based on the artificial vagina: securing bulls' active and reliable participation in breeding became more rather than less important as a.i. spread and bulls disappeared from farms. The new technology demanded a reliable user, since increasingly fewer bulls had to serve a growing system built up to distribute semen, impregnate cows and keep the milk flowing. Therefore, while a.i. eliminated the link between pregnancies and bulls serving cows, it placed the act of service as such in greater focus than before. In Sweden, this led to a centering of bulls' sexual agency and ability, and to the construction of a particular form of veterinary expertise to evaluate these factors. Third and finally, the centering of bulls in a.i. also formed the basis of an attack on the prevailing system of Swedish pedigree breeding. Veterinarians began to argue that the systematic and recurring types of problem that the bulls exhibited when serving in the artificial vagina suggested that there was something fundamentally wrong with the way that they were bred. This argument contributed significantly, I contend, to the rapid replacement of the established, exterior-and ancestry-focused breeding system with a quantitative, more directly production-oriented breeding system in the 1950s.
Ultimately, then, this paper argues that the dynamics of human-animal interactions in early a.i. not only helped shape this crucial technology itself but also veterinary expertise, the institutions of breeding, and parts of the agrarian economy. These things to some extent became co(w)-produced: formed by bulls' behaviors and bodies as humans interacted with them to gather semen for artificial insemination.

Background: A.I. technologies and initial A.I. Experiments in Sweden
The first mention of a.i. as a usable breeding method in the Swedish trade press dates to 1930. 14 Systematic attempts to use it began in 1935. During an eight-year experimental program, its techniques stabilized, and it became strongly associated with the veterinary profession. As a background to my main analysis, which considers the period after 1943 when a.i. was widely introduced in farming, this section will present the technologies involved and discuss the experimental program that established a.i. in Sweden.

The technologies and purpose of early A.I.
A.i. in cattle builds on three different sets of technologies: those used to gather semen, those used to prepare the semen for insemination, and those used to inseminate. 15 In 1935, none had taken on fixed forms. Inseminations were performed with a syringe and a glass tube. The initial method was to use a vaginal speculum, by which the inseminator, using a headlamp, could see the location of the inseminating tube in the reproductive tract. 16 Later, an alternative to this vaginal method was developed, which built on inseminating by feel instead of by sight. The inseminator then used a hand inserted into the rectum to guide the instrument. By the 1940s, this recto-vaginal method had become predominant. 17 Before inseminations could be performed, the semen was diluted with agents that served to conserve and extend it, to keep it viable for a few days and so that a single ejaculate could be used to inseminate a large number of cows. Considerable experimentation took place to identify the best dilution fluids and most appropriate concentrations. 18 From the 1950s, deep-freezing of semen was also developed, but in Sweden, this method was only adopted after the end of my study.
The problem of gathering semen had a diverse set of solutions: a 1936 veterinary introduction to a.i. listed seven different techniques. Three depended on the natural service of a cow, with semen either spooned out from the vagina afterwards or collected in a sponge or container placed in it beforehand -or, alternatively, the bull could be made to wear a condom. Two did not require service, but produced ejaculation by way of massage or electrical stimulation. Finally, semen could be collected using an instrument called an artificial vagina ( Figure 1). This became the most widely used method and is in focus here. 19 The artificial vagina was -and is -a rubber-lined tube heated and pressurized by warm water, meant to simulate the sensation of a real vagina. Its use built on harnessing normal mating behavior. Although several factors can contribute to sexual arousal in bulls, visual cues are the most important. Once interested in a cow, the bull will approach to gather her receptivity. If she stands still (standing still to be mounted being an important sign of estrus), the bull is likely to initiate the mating act. He will get an erection and then mountclimb up onto the cow -and start moving his penis back-and-forth, seeking contact with the vulva. Upon such contact, an ejaculation reflex is triggered: the bull will thrust -jump forward -and ejaculate in the vagina. 20 When using the artificial vagina, bulls were similarly allowed to mount a 'teaser' animal (first, cows in heat were used, but it was soon found that any cow or even another bull could be used, as long as the animal could not move away). When the bull had mounted and began performing seeking movements of the penis, the human collector would grab the bull's sheath and steer the penis towards the lubricated artificial vagina. Ideally, the instrument would then trigger the ejaculation reflex, with the semen being gathered in a test tube attached to the device. 21 Like the spoon, container, and condom methods, gathering semen in the artificial vagina depended on bulls playing an active role. Unlike those methods, however, it also involved an active role for a human. Gathering semen in the artificial vagina built on an intimate form of human-animal interaction in which the human party had to be very attentive to the movements and sexual behavior of the bull.

A.I. at the Institute for Livestock Breeding
Not only was the instrumental setup of a.i. in flux in the mid-1930s, but its ultimate purpose was not settled. In the Soviet Union, which had pioneered the practical application of a.i. in cattle, it was used primarily in response to a general lack of quality bulls. Other recognized advantages were that a.i. could save money and labor, and increase occupational safety by reducing the need for bull handling. Furthermore, it could be used to negate spatial limits to breeding (it was much easier to transport semen than livestock over long distances), to fight sexually transmitted diseases, and to overcome certain forms of infertility. Finally, some scientists wanted to use a.i. to develop a system for the progeny testing of bulls: of determining their breeding value on the basis of their daughters' production.
It was for this latter reason that experiments with a.i. had started in Sweden. This was linked to the beneficial conditions for livestock genetics research that existed at the semiprivate Institute for Livestock Breeding at Wiad outside Stockholm. Founded in 1928 in response to international advances in livestock genetics and breeding research and with the help of a private donation from the Knut and Alice Wallenberg Foundation, its remit was to conduct 'scientific and practical studies' to 'improve the characteristics of the domestic animals from an economic point of view'. Geneticist Gert Bonnier from the University College of Stockholm, who had a prior interest in the genetics of livestock breeding, was appointed its first director. Zoologist Kåre Bäckström headed the institute's laboratory work. 22 Bonnier's principal preoccupation when it came to cattle breeding was with progeny testing, the identification of good sires based on their daughters' production. Understood by livestock geneticists as key to a rational selection -since it allowed for direct judgements about inherited milk-production traits -progeny testing had proven difficult to implement since cattle procreate slowly and have only one calf at a time. A.i., Bonnier realized in the early 1930s, could be a way to overcome this difficulty, as it made it possible to breed larger number of cows from a single bull and thus could expand the basis for the evaluation. He sent Bäckström to Britain and the Soviet Union to learn the method, and from 1935, a program of a.i. experimentation began at Wiad, working with the cows of neighboring farmers. Within a year, the institute reported that it had mastered the fundamentals of the method. 23 Bonnier's work with progeny testing was not merely intended as a scientific experiment but also as an intervention into the organization and economics of cattle breeding in Sweden. In the interwar period, organized breeding 24 was dominated by well-off farmers working according to the principles of purebred breeding, with public pedigreekeeping and selection by ancestry. 25 They had defined two principal dairy breeds, Swedish Red-and-White Cattle (Svensk röd och vit boskap, SRB) and Swedish Friesian Cattle (Svensk låglandsboskap, SLB) and sought to develop these breeds in conformation with their specified exterior standards and production criteria. The administrative work of registering animals and organizing breeding was the purview of breed-specific pedigree-breeding associations, organizations dominated by commercial breeders who bred and raised breeding stock for sale. Overseen by the government, the pedigree-breeding associations kept herd books (public registers and genealogical records of breed animals) and arranged transactions of pedigree stock, most importantly by organizing public auctions where the most valuable breeding bulls were sold and breed standards and norms therefore reproduced. Although milk production records had been incorporated in this system after milk recording began in the late nineteenth century, it remained dominated by judging how well animal exteriors conformed to the established breed type up until World War II. 26 This was what Bonnier's development of a progeny testing method, exclusively focused on milk and butterfat yields, was meant to challenge.
Neither Bonnier nor Bäckström originally saw any need for veterinary involvement in their a.i. work. But the Wiad project was nevertheless taken over by vets, overseen at a remove by the Veterinary College's professor of obstetrics and ruminant medicine, Nils Lagerlöf. 27 A.i. became coded as a primarily veterinary technology, which in turn gave the profession significant influence over Swedish a.i., including over the subsequent training of all a.i. personnel. By 1943, when a.i. began to be implemented in farming, its technical aspects were wholly under veterinary control, and the emerging cooperative a.i.-breeding associations hired vets for all a.i. work. Its scaling-up in the context of practical farming brought new problems, however, as an increasing number of bulls struggled with using the artificial vagina. It is to these troubles I now turn.

A history of failure: the first years of A.I. in Swedish farming
Already at Wiad it had become apparent that semen of sufficient quality could not always be obtained, as many bulls failed to adapt to the demands of the new technology. Some had a low sex drive and took too long to initiate mating, preventing an efficient gathering process. Others did not thrust or ejaculate in the artificial vagina, or if they did, would not deliver semen of sufficient quality for a.i. Some hurt themselves when trying and subsequently refused further attempts. Yet others performed the act successfully but would then start to behave so aggressively that further semen gathering had to be abandoned. When a.i. expanded beyond Wiad from 1943, these problems grew more frequent. Their pervasiveness, I show here, was the principal issue that early Swedish a.i. had to deal with.

'No one is fooled': the problem of identifying bulls for early A.I.
From the outset, everyone involved with a.i. recognized that gathering semen was a complicated process that required use of the correct technique as well as a good sense of how to interact with a large and potentially unpredictable and dangerous animal. Inexperienced gatherers often failed to get bulls to cooperate at all. In October 1942, vet Eric Andersson, who only had received some desk-based instruction in a.i., wrote to professor Lagerlöf and asked for advice as he could not get bulls to serve in the artificial vagina. 'I', Andersson wrote, 'have tried both young and old ones, but no one is fooled, even though I have greased the [artificial] vagina with fat and mucus from cows in heat and in spite of warm water between the rubber layers. There must be a particular technique that I do not know'. Lagerlöf replied that 'it is relatively important both to have the right temperature for the . . . water and to use the correct tension, so that the friction between penis and vagina is suitable for the bull', but also underlined that 'it is particularly important that one uses a technique that works completely without problems'. 28 This was not only a matter of 'fooling' the bulls into ejaculating but also one of their health and safety, since strong forces were in play in the bovine mating act and injury to the penis was known to sometimes happen if the person maneuvering the vagina did not have sufficient skill. 29 But the problems were more profound than a lack of training. Experiences both from Wiad and from Denmark, where a.i. had been in regular use since 1936, suggested that many older bulls simply refused service in the artificial vagina, no matter the technique and experience of their human interactant. Even if they did serve, the semen would not always be of sufficient volume or concentration for a.i. When Folmer Nielsen, professor of obstetrics at the Danish Agricultural and Veterinary University, lectured in Stockholm in 1944, he cautioned that both young, over-eager bulls and old, slow ones could be difficult to use for a.i. 30 In spite of such cautions, and some early controversy over a.i'.s potential risks for future sexual behavior and libido, many farmers enthusiastically welcomed the new technology. 31 They appreciated its labor-saving and safety-enhancing properties, and saw it as a way to access premium-quality bulls. More progressive farmers who were invested in large-scale production also recognized the potential of using a.i. to intensively breed progeny-tested bulls. 32 The number of cooperative a.i.-breeding associations grew quickly in south and central Sweden between 1943 and 1945, and a.i'.s chief proponents formed the National Association of Swedish A.I.-Breeding Associations (Riksorganisationen Sveriges Seminföreningar, RSS). 33 Yet a review of records from one of the first a.i.-breeding associations formed in Sweden, the Eskilstuna elite bull association, confirms that Nielsen's warnings were borne out in farmers' early experiences. From its constitution in November 1943 and through all of 1944, the Eskilstuna association repeatedly but often unsuccessfully tried to find bulls that would function in an a.i. setting, leading to a tragi-comical series of acquisitions, loans, and losses. The first bull acquired, 367 Hagel-Hero, performed satisfactorily. The second, 73 Wärnsta-Hero, promptly broke a femur and was removed before having been evaluated. This led to increased strain on the third bull, 192 Wärnsta-Hero, whose performance and semen quality quickly worsened. The association's board of directors nevertheless felt they had to let him stay on, since at that point, 367 Hagel-Hero was also exhibiting signs of exhaustion. Two more bulls were then borrowed for a brief period, but neither proved to deliver sufficiently high-quality semen. Another two were inspected, but the association declined to acquire them since they, too, produced poor semen. Instead, a fifth bull, 94 Lindenäs, was bought, and later yet another two, the young 202 Lindenäs and 556 Getinge. Unfortunately, both 556 Getinge and 94 Lindenäs often refused to 'thrust in the artificial vagina'. Finally, in late 1944 an eight and a ninth bull were bought and both proved to meet the association's needs. 34

Interpreting the semen supply problem
For those in leadership positions in RSS and the new a.i. associations, the problem of finding bulls that would reliably serve in the artificial vagina and deliver a.i.-grade semen was not just about what we might call a semen supply problem. Being breeders, they were concerned with individuals, and in this case specifically with using progeny testing to identify individuals with particularly good traits for milk production. Their planned breeding program could not be realized unless those best individuals could be trusted to deliver semen regularly. The problem was exacerbated by the fact that bulls would be relatively old (approximately 6 years) before the progeny analysis could be completed, and older bulls had particular difficulties with a.i. Taking stock of the situation in early 1946, vet Ivar Dyrendahl (a student of Lagerlöf's) noted that only a small number of bulls reach [the age required for progeny testing] and it is . . . difficult at that point to introduce them in a.i. For example, older bulls often demonstrate an aversion to serving in the artificial vagina, and often the fertility of their sperm declines. . . with increasing age. 35 To vets like Dyrendahl and Lagerlöf, the problem had another and more serious dimension too. There was a long-standing conviction among Swedish vets that bovine fertility problems were to large extent hereditary. Since the early 1930s, Lagerlöf had forcefully argued against the use of animals for breeding if they demonstrated reduced fertility that was not readily explainable by environmental factors. And not only did the prevalence of problems in a.i. seem to confirm concerns about the degeneration of reproductive fitness, the vets also understood a.i. itself to pose a new and potentially grave threat, since it sped up breeding and could enable the degenerative traits to spread quickly within entire breeds. 36 Lagerlöf, for whom this matter was a principal research interest, therefore increasingly emphasized the importance of removing weak bulls. 'Of all preventive work against infertility', he argued in 1944, 'the breeding hygienic measures are the most important, that is, that we if possible seek to prevent that the animals are born with the predisposition for infertility'. This was particularly important, he noted, for bulls intended for a.i. associations. 37 RSS and veterinary leaders thus shared the view that it was necessary to find a way to systematically weed out sexually weak bulls before they were put to use in a.i. associations. An a.i. bull had to tolerate the artificial vagina well and demonstrate sufficient sexual strength that he posed little risk of spreading infertility-related traits. Vets therefore begun to develop ways of evaluating these characteristics.

Veterinary evaluations of sexual fitness
From 1945, RSS enlisted Lagerlöf and some of his practicing colleagues to judge the a.i. suitability of all young bulls offered at the pedigree-breeding associations' auctions that were of interest to the a.i. associations. The evaluation consisted of an observation of the bull's behavior when using the artificial vagina and a laboratory analysis of its semen. 38 Experience soon showed that the first part was very revealing in itself, provided that one had seen enough bulls and could recognize recurring behavioral patterns. Good judgments required thick descriptions of the relevant behaviors: in March 1947, Lagerlöf instructed a colleague to 'rather write too much than too little' on the form provided, and 'rather be too critical than the opposite'. 39 The judgements focused on bulls' behavior in the artificial service situation and on how they tolerated the intimate interaction with a human that it implied. Many bulls were untroubled by having their penises steered towards the artificial vagina in the midst of the mating act, but some reacted very negatively. For example, Lagerlöf's student Bengt Lundgren reported in 1945 about 102 Leregårds-Hero that 'after the first jump, he was very suspicious of me' and that he would thereafter not mount the cow unless the vet was behind him. 40 A more colorful anecdote is reproduced in a retrospective account by another of Lagerlöf's students, Thure Swensson, as a way of illustrating the complex interplay between the bull, the bull handler, and the vet that semen gathering required. On the occasion that Swensson recalls, 'the bull was dissatisfied after performing the service and turned towards the veterinarian so forcefully that his handler could not hold him back. The vet had to run, followed by the bull and last the handler hanging on to the rope. There was a wild rush through the barn until the bull found his own box with his comrades nearby'. 41 While the tone is humorous, it was presumably a situation that could have ended in a fatality.
During two years of evaluating bulls on RSS' behalf, the involved vets built up a set of shared standards about what had to be expected from them. When Lagerlöf spoke at RSS' 1947 annual meeting, he briefly summarized these. An a.i. bull had to 'serve impeccably in the artificial vagina', Lagerlöf said, 'with no hesitation and a forceful thrust. He shall of course be given time to gather himself before mounting, but shall not be allowed to hang onto the cow with weak thrusting, nor be overly cautious with his sheath when [the vet] grabs it. When thrusting, he also has to show that he can arch his back in the right way'. 42 Lagerlöf's correspondence suggests that many Swedish vets likewise developed a sensitivity to the different behaviors that problematic bulls manifested in an a.i. context, whether they were involved in formal evaluations or not. In a November 1947 letter, vet Sigurd Dahlqvist with the Örebro a.i. association told Lagerlöf about troubles he was having with 89 Brobygårds-Hero, one of the association's bulls. He was, Dahlqvist wrote, of the 'sewing machine type', referring to a recognizable type of behavior that the bull exhibited. He would not move determinedly to mount the teaser cow, but would 'crawl up to her' before getting a partial erection. He would then mount but fail to perform a distinct thrust, only making 'a lot of short little thrusts' before dismounting. It was rare for Dahlqvist or his colleagues to secure an ejaculate. The best chance to get one seemed to be if they could 'catch him in action without touching the sheath', but that technique made it 'pretty difficult to hit the spot', that is, getting the penis into the artificial vagina. Similar coordination problems -though in this case not related to the bull -could also arise when using cows not in heat as teaser animals. If they could not be properly fixated, they would often be restless and 'jump back and forth', making it more difficult to obtain the sample. 43 By this time, it was broadly recognized not only that many bulls across the country were refusing to conform to the 'script' of the artificial vagina, but that they did so in particular ways, and that the problems were especially prevalent in the SRB Red-and-White breed. 44 Among the bulls that Lagerlöf and his colleagues had evaluated for RSSall expensive young bulls from more or less renowned breeders -only 30 percent passed the examination. 45 Few of those disqualified were wholly unable to deliver semen, but they failed to meet what the vets considered necessary standards of mating behavior or semen quality to be suitable for a.i. Before a.i'.s introduction, such rejections of pedigree bulls on fertility grounds had been rare unless the bull was completely sterile or wholly impotent. Essentially, this shows that the demands of the new technology were reframing the way reproductive behavior was understood. With new and more detailed attention paid to precisely how bulls performed the mating act, things like 'sewing machine' performances became both evident and causes for rejection. The introduction of a.i, in other words, was influencing both bovine behavior and the veterinary interpretation of such behavior. In the next section, we will see how this interpretation became more nuanced as the profession, in part also responding to insurance interests, focused more closely on evaluating bulls for semen supply through the 1940s.

'Without encouragement or urging': A.I. and the centering of bull agency
The veterinary tests performed on RSS' behalf also attracted the attention of the Swedish livestock insurance companies. Since 1928, these had offered insurance against infertility in breeding bulls, and with time they had taken on considerable infertility risk. The rising prevalence of impotence and infertility in the context of a.i. was therefore costly for them, and they increasingly pushed for the veterinary examination of sexual fitness to be extended to all pedigree bulls for sale at auctions, with the results made publically available. This stimulated a further development of the veterinary evaluation methods. They grew, this section argues, more elaborate as well as more concerned with the agency of the bull in the mating situation. Trying to 'fool' the bull into serving was increasingly seen as inappropriate, with vets instead attending closely to how it behaved on its own accord and without external stimuli.

Insurance companies and the evaluation of breeding potential
The insurance companies' intervention had initially met resistance from the pedigreebreeding associations. These represented the sellers' interests, and commercial breeders had a natural reason to dislike stricter restrictions on the insurability of bulls. The leverage of the insurers, backed up by veterinary expertise, was nevertheless significant enough that an experimental evaluation system was developed and put into effect from 1948. The most prominent of the insurance companies, the Scandinavian Livestock Insurance Company, also financed scientific research on different factors' influence on mating behavior and semen quality, using twin bulls as experimental subjects. The veterinary part of these studies was carried out by Nils Lagerlöf's student and protégé, Allan Bane, who had been involved in the experimental a.i. work at Wiad and had taught a.i. at the Veterinary College since 1944. 46 Bane's work contributed to a certain standardization of how vets judged mating behavior. Earlier, the evaluations had been ad hoc, with different vets sometimes emphasizing different aspects. Bane, however, was performing a scientific study and had to develop a framework that would allow for valid comparisons. In 1950, he circulated a memo presenting his method in detail. 47 It required the vet to gather at least two semen samples in the artificial vagina and then -since some aspects of mating behavior were impossible to observe while handling the instrument -to let the bull serve naturally once or twice. A cow that was not in heat should be used as the teaser animal. Crucially, Bane also stressed that the evaluation should be predicated on the bull's own desire to mate. Special interventions to excite him 'should under no circumstances be allowed'. He was to 'handle himself freely and perform the service when he so wishes without encouragement or urging'.
Bane's method broke the mating act down into four phases, to be judged separately as far as possible. The first was the erection. Some erectile problems, such as complete impotence, were easy enough to diagnose. Others were more subjective and only apparent in intimate interaction. Bane noted, for example, that some bulls had a seemingly normal erection, but when using the artificial vagina, the vet would be able to feel that 'the penis [was] not normally erect and stiff but flexible and limp and that the thrust [was] weak and without real power'. The second point of judgement dealt with the manner in which the bull mounted the cow and how he carried his body while doing so. It called for attention to the slightest details of posture. Vets should be especially watchful, Bane cautioned, of bulls that were unable to get sufficiently far up on the cow. These generally struggled to complete natural service but could sometimes perform well in the artificial vagina. They were nevertheless to be deemed sexually weak, with a poor prognosis. The third moment to be evaluated was the seeking movements of the penis that the bull performed after mounting. Problematic bulls often demonstrated a short protrusion of the penis and overly quick seeking movements. They could remain on the cow for a relatively long time, though often without successfully serving. The quick seeking movements also made it very difficult for these bulls to use the artificial vagina, that is, for the human semen collector to coordinate his movements with theirs. The fourth step was the final thrust and ejaculation. This could be particularly problematic when serving artificially. Certain bulls did not ejaculate in the artificial vagina, but simply continued to perform 'the seeking penis movements . . . more or less deeply in [it]'. Other bulls did thrust, but weakly, providing a low-quality ejaculate. The pressure and temperature of the artificial vagina could be adjusted to try to overcome these problems, but that only helped some of the struggling bulls (and given their preoccupation with heredity, vets would be disapproving of bulls needing special adjustments anyway).
Bane's instructions provide detailed insights into how vets tried to analyze bulls' sexual behavior in the context of a.i., and demonstrate how use of the new technology was negotiated in complex human-animal interactions. With the artificial vagina, new animal behaviors arose that could be normatively judged by the vets. We saw one example of judgements about proper and improper users in Lagerlöf's 1947 address to RSS above. Bane's instruction implied a similar evaluation, but was more detailed and nuanced. A proper user, his memo implied, was a bull who needed no other stimuli than seeing a cow to get aroused, who mounted determinedly with the correct posture and a full erection, who performed strong but measured searching movements, who tolerated the steering of his penis towards the artificial vagina, and who immediately thrusted and ejaculated upon contact with it. An improper user would fail to meet these standards in one or more ways.

Institutionalizing veterinary expertise
When these understandings of different kinds of use and users of a.i. began to ground practice, they had far-reaching consequences: through examinations like the one prescribed in Bane's memo, the entire system of organized cattle breeding in Sweden began to shift. From 1950, the livestock insurance companies announced that the science was now far enough advanced for them to demand that all bulls to be insured had passed a sexual fitness examination. 48 This meant that from then on, young bulls that failed to satisfy veterinary demands for sexual strength had limited market value, even if they would have been very valuable according to the established breeding norms of exterior characteristics and ancestry.
Demands from buyers of a.i. bulls and, later, from the insurance companies, generated paperwork that further help us understand the vets' judgements about use and users of the artificial vagina. Most notably, Bane's emphasis on the subjectivity to the bulls resound throughout the evaluation reports. As we saw above, they sometimes describe bulls directly demonstrating their displeasure. 451 Hamra-Hero, whom Bane himself evaluated in September 1951, was 'very suspicious'. Bane reported how he 'turned the horns toward me and was very embarrassed and attentive to me while mounting', not appreciating the close human contact. In this case, however, a short rest and some care and a good 'pat' calmed Hamra-Hero down and he then served perfectly in the artificial vagina. 49 Even when bulls did not respond aggressively to the semen gathering, vets always ascribed other forms of subjectivity, intentionality, and sentience to them. Every bull was rated on his 'willingness' or 'eagerness' to serve, terms that served as proxies for general libido. Bulls that disliked having their sheath touched and their penis steered towards the artificial vagina were deemed 'embarrassed' whereas those who did not react to the human touch were considered 'unembarrassed'. The reports reproduced in Figure 2 and Figure 3, of tests performed on behalf of the SRB breeding association, exemplify a negative and a positive evaluation, respectively. From the report in Figure 2, we learn that 61 Östuna-Hero was 'sluggish' ('trög') and 'unwilling' ('ovillig') during his test, that he performed a 'short thrust' ('kort stöt'), demonstrated 'apathetic conduct of the penis' ('stum penisföring') when seeking, and that he was 'embarrassed' ('generad') with the artificial vagina. In contrast, Figure 3's report tells us that 321 Boxholm-Hero was 'quick' ('snabb'), 'willing' ('villig'), and 'relatively calm' ('rel. lugn'), and that he performed a 'very good thrust' ('mkt god stöt') with 'very good protrusion' ('mkt god utskaftning') of the penis. 50 Over the following years, the vets involved in a.i. developed the testing protocol further. At a 1953 conference for a.i. vets in managerial roles, Carl Axel Hultnäs, then employed by the Red-and-White Cattle pedigree-breeding association to evaluate all bulls proposed for sale at its auctions, presented an extension of Bane's system. Hultnäs wanted to evaluate six rather than four phases: general libido, erection during seeking, protrusion of the penis, the manner of seeking movements, thrust, and the penis protrusion at the moment of thrusting. He also proposed a system of category-based ratings for each phase that would allow for easier comparison and be more compatible with the punch-card system of data management that was being introduced in Swedish a. i. The categories suggest the wide range of recognizably different behaviors that the vets thought they could identify: a bull's general libido, for example, was to be assigned to one of seven categories: 'no interest at all' in the cow; 'limited interest . . . no attempts to mount'; 'limited interest . . . apathetic attempts'; 'moderate interest, mounting after hesitation'; 'relatively willing; mounting without strong drive'; 'proceeds quickly; mounts after good preparation'; 'hard [libido]; wants to mount quickly'; and finally 'hot to completely thunderous sex drive'. 51 The system was still predicated on letting the bull handle himself: it was his subjective interest and capabilities that were to be evaluated. Hultnäs' continued interest in the sexual behavior in bulls led to him defending a PhD dissertation on variations in mating behavior and semen picture in young bulls in 1959. In the dissertation, he added a seventh category of evaluation ('Attitude of the body in mounting, seeking, and thrusting') while also acknowledging that '[s]ervice is a complex act with no distinct limits between the different phases'. 52

Centering bull subjectivity
These examples of how vets interacted with bulls when gathering semen, and of how the veterinary expertise in evaluating service became institutionalized, continue to show how the particular performance of the mating act took on a new importance in the a.i. context. They also show that judging this performance came to be predicated on what was seen to be the bulls' own preferences. Letting them act freely-'without encouragement or urging'-was central to judging their suitability for a.i. Vets therefore focused on what they deemed to be expressions of bulls' intentionality and agency in the mating situation.
This problematizes some received understandings of a.i. It is associated, as we saw in the introduction above, with the industrialization of reproduction, and a considerable strand of critique of industrial farming suggests that it builds on subjugating animals to the demands of technology. For example, in a recent discussion of Swedish dairy advertising, sociologist Tobias Linné explicitly uses a.i. as one example of how 'animals are incorporated into production technologies' and are 'modified and designed to suit the production system and optimize productivity . . . are de-animalized, alienated from their own bodies and from their bodily functions'. 53 This is not, I argue, what happened in Swedish a.i. in the 1940s and 1950s. Turning semen into just another farm input was a fraught and difficult process that to a considerable degree came to be constructed around the agency of the involved bulls.
To Linné and other critics, my objection might seem to miss their point: no matter how specific husbandry technologies were organized, they still ushered in a new production system that radically worsened the conditions under which farm animals lived and died. Nevertheless, their rhetoric normally assumes that industrial farming has only ever been about the imposition of new technologies, leading to the complete instrumentalization and alienation of the animals. This was not the case for bulls in Swedish a.i., even though highly instrumentalizing technological alternatives in fact existed, including devices for producing ejaculations through electric stimulation, massage-based techniques, and the surgical creation of a fistula through which to extract semen. In the early and mid-1950s, there was an active discussion in the international literature of electroejaculation as a potentially useful method to gather semen from bulls that could not or would not serve in the artificial vagina. 54 In Sweden, however, such methods were never posited as potential solutions to the recruitment problems that defined early a.i, and international experiments with them were met with skepticism from both 'humanitarian and health viewpoints'. 55 Rather than bypassing the artificial vagina when bulls would not use it, the focus in Sweden thus remained squarely on finding suitable users. This stance led Swedish vets and a.i. breeders to resist desubjectification of a.i. bulls. Through the constant use of terms like 'willingness', 'eagerness' and 'duty', they instead framed bulls' subjectivity in the mating situation as key to the successful development of a.i. This finding evokes a point made earlier by Abigail Woods about modern pig production: that received accounts of modern agriculture are mistaken in conceiving industrial farming as being solely about domination. Both experts and farmers, Woods contends, 'perceived pigs as active, sentient individuals' and believed it was essential, also from a commercial point of view, to construct farming systems that respected their natures. 56 Similarly, Swedish a.i. vets thought it crucial to not force a.i. on bulls, however valuable they otherwise were, but to only use those naturally at ease with the artificial vagina. To do otherwise, they believed, would be to undermine cattle breeding as a rational pursuit.
The flip side, of course, was that it was easier than ever before for bulls uncomfortable with the new technology to be deemed 'unwilling', 'uneager', or 'embarrassed'. While such veterinary ascriptions of intentionality, agency, and perhaps even understanding to the bulls served human purposes far removed from their lived realities, the veterinary judgements are, nevertheless, valid sources both of systematic differences in mating behavior and a new emphasis on mating behavior from human observers. Such emphasis was a wholly new form of evaluation closely integrated with a.i. Earlier, there had been little concern with the details of how the bull mated, as long as it did mate, or could eventually be coaxed into doing so. This changed with the introduction of the artificial vagina and a.i'.s demand for reliable semen producers. In the next and final section, I will go on to show how this centering of the bovine user had broader implications for organized cattle breeding in Sweden.

'Ferdinandish' bulls and the decline of pedigree breeding
In an earlier paper, I have argued that Swedish veterinary reproductive science's insistence on seeking hereditary explanations for bovine fertility issues helped reshape postwar cattle breeding in the country. 57 The argument of this section overlaps to some extent with the arguments I make there. However, as the earlier paper focuses on the long-term trajectory of the scientific development, it does not engage with the nature of these disturbances as they related to the introduction of a.i. Here, I therefore complement and extend my earlier argument by showing how the crucial veterinary arguments, and the shifts in the breeding system that they helped bring about, hinged specifically on the artificial vagina.

Tensions around pedigree breeding
From the late nineteenth century, systematic cattle breeding had been initiated in Sweden as part of a general drive towards the rationalization of cattle husbandry, driven by rising dairy prices. As I discussed above, by the interwar period this organized pedigreebreeding system was dominated by well-off farmers who produced purebred animals for sale. The market value of a breeding bull was determined by its ancestry and, particularly, by its exterior conformation to breed norms. Both the two main cattle breeds were primarily understood as dairy breeds -with the SLB breed being more pronouncedly specialized -but meat remained an important byproduct of dairy.
Although the pedigree-breeders' systematic approach represented a modern break with earlier husbandry practices, in the 1930s and 1940s, it was itself challenged by more radically modern arguments for a scientific breeding, unconcerned with animal exteriors and geared solely to productive efficiency. As noted, the attempt to use a.i. to introduce progeny testing in Sweden was part of this challenge. Its proponents often mounted direct attacks on exterior-focused breeding, as when Gert Bonnier discussed principles for animal evaluation at the Royal Academy of Agriculture in 1932 and noted that he considered 'the significant consideration of exteriors' in breeding both 'wrong and probably damaging'. 58 But in 1932, Bonnier did not speak from a position of strength.
Apart from a limited group of scientists, progressive breeders, and radical debaters, most people in farming strongly believed that how cattle looked was an important aspect of breeding and husbandry. A well-conforming breeding herd with prominent lineage attracted significant attention and afforded both high status and monetary return to its owner. 59 Norms about how animals should look were disseminated not only through bull auctions but also other forms of animal shows as well as through the trade press and in higher and lower agricultural education. It was this entire system of exterior-focused breeding that the veterinary interpretation of the problems with early a.i. began to undermine.

Exterior breeding and the secondary sex characteristics debate
From the late 1940s, vets began to argue that the sexual problems apparent in relation to the artificial vagina required a redefinition of the value of individual breeding animals and lineages. Not only did they see sexual problems as largely inherited, they also thought there were reasons to believe that the problems were directly related to breeders' priorities. The crux of the theory, most forcefully put forward by Lagerlöf, was that the focus on breeding for particular exteriors had led breeders towards trying to remove the secondary sex characteristics of the bull, and that this had interfered with bulls' hormonal make-up and sexual development. These endocrine issues affected their temperament, making them, as he argued in 1950, 'sluggish and slow'. 60 Sluggish, as we have seen, was veterinary terminology for bulls that were hesitant or reluctant to serve, so Lagerlöf's argument linked what he saw as problematic breeders' practices with the veterinary experiences of evaluating bulls with an emphasis on their sexual agency. By extension, this meant that the breeders were to blame for the widespread failures in using the artificial vagina.
Here is a more extensive take on the problem that Lagerlöf presented to the Sixth Annual Conference of the American Society for the Study of Sterility in San Francisco in June 1950: In judging these breeding problems we must try to consider the matter also from a biologic point of view, and we must caution the breeders firmly against eliminating the secondary masculine sexual characteristics in the bulls. In some breeds the cattle breeders have attempted to produce a so-called purebred type of bull. Consideration has then been paid only to external characteristics and bulls of a feminine type have been preferred. Thus, in their efforts to produce a purebred type these breeders have bred bulls which are too refined and they have eliminated the secondary masculine sexual characteristics. In certain breeds the result has been that the bulls to a great extent have become distinctly of a feminine type, have become very good-natured, and often show unwillingness and poor ability to serve. 61 Again, he emphasized how breeding practices had led to large numbers of bulls being uninterested in serving. Some American newspapers happily seized on this point, proclaiming that a 'Ferdinand brand of bull' (referencing a well-known Walt Disney animated short film about an unusually calm and withdrawn bull) had emerged from Swedish breeding and that these 'Ferdinandish' bulls were too high bred to 'be bothered' anymore. 62 The veterinary arguments about the bulls' masculinity and femininity in important respects parallel Gabriel Rosenberg's discussion of pig breeding in turn-of-the-century United States. 63 The American pig breeders, too, were highly focused on the masculinity of their boars and linked this both to their sexual prowess and their capacity to pass on their traits to the next generation. The preoccupation with masculine characteristics therefore perhaps reflected older and broader gendered views of virility and vigor. But there is an important difference: Rosenberg's pig breeders went to great lengths to ensure that hogs would breed even when they resisted doing so, while the Swedish vets' understanding of low sex drive as a medical and endocrinological problem led them to the opposite conclusions: bulls that did not want to serve should not be used at all.
Another important line of scientific work on bull infertility in Sweden at the time was Allan Bane's aforementioned insurance company-funded experimental studies of mating behavior and semen quality in bull twins. Using twins made it possible to isolate environmental from genetic influences, and Bane claimed to have proven that both mating behavior and semen quality were indeed inheritable characteristics. 64 As we saw above, his studies built on gathering semen in the artificial vagina and on evaluating bull's behavior as he did, making the vagina both the reason for the study and the tool that made it possible.
Both Lagerlöf's worry over 'feminine' and 'sluggish' bulls and Bane's studies on the heritability of mating behavior and semen quality thus drew directly on the evaluation of minute details as bulls served, naturally and in the artificial vagina. Lagerlöf's idea about the 'Ferdinandish' bulls was particularly important. It circulated in the agricultural press, and also attracted considerable attention after Lagerlöf used it when acting as an expert witness in a highly publicized court case involving the sale of an allegedly sterile bull. 65 Taken together, these veterinary arguments contributed significantly to weaken the position of those breeders who preferred exterior-focused breeding. In 1951, the postulated link between exterior breeding and reproductive capabilities had become salient enough in breeding debates to be addressed by the outgoing chairman of the SRB breeding association, estate owner Gustaf von Hofsten, at the association's annual meeting. Although von Hofsten defended the association against accusations that it had 'worked too much on improving the animal's exteriors', the proof he offered to support his claim that the association sought to promote a 'healthy breed' and not just an aesthetic one was its introduction of 'careful evaluations of the reproductive capabilities of the breeding bulls'. By thus implicitly accepting that infertility was a health problem linked to exterior breeding, von Hofsten affirmed Lagerlöf's thesis. 66 The new evaluations that von Hofsten mentioned had been introduced in response to the new insurance regulations. The SRB association had always performed preinspections of bulls to be sold at its auctions. Earlier, however, this inspection had been carried out by an officer of the association who was also an experienced breeder. He would rate the proposed bulls' exteriors, refusing to admit to the auction a bull that deviated too much from defined breed norms. With the new focus on mating behavior, however, this pre-inspection was instead carried out by a veterinarian and focused solely on determining the sexual fitness of the bull through the use of the artificial vagina. 67 This replacement of long-established aesthetic judgments striving for the type-beautiful with a strictly functional test took place in the heart of pedigree breeding and was thus symbolic of a more general shift away from emphasizing well-conforming animals. Instead, and as I have argued in my earlier paper, a new group of a.i. breeders and livestock geneticists began to control the direction of dairy cattle breeding. They were largely uninterested in conformation, looking instead at cattle as bearers of quantifiable and optimizable traits for milk production volumes and butterfat levels. 68 This was a development in line with the postwar emphasis on rational farming, with little place for older ideals about well-fed and well-bred animals communicating social status.
This quantification of breeding was -as is well known -made possible by a.i. and the progeny testing it enabled. The novelty of my finding is not that a.i. changed breeding practices in Sweden, but that it brought about a redefinition of breeding values to also include sexual performance, and that this in turn opened up for a quick transformation of the breeding system as a whole. This redefinition was made possible by the systematic use of the artificial vagina to evaluate breeding bulls and by the veterinary centering of bull agency in such evaluations. By bringing out the 'sluggishness', 'short thrusts', and 'apathetic penis conduct' that many otherwise celebrated bulls manifested when using the artificial vagina, and arguing for these being hereditary problems for which their breeders bore a heavy responsibility, they undermined the legitimacy of the established breeding system. This opened up conditions for the rapid transformation that took place from the late 1950s and on.

Concluding discussion
In this paper, I have shown how the organization and context of Swedish a.i. developed together with its human and animal use. More specifically, I have made three points: first, I have shown that early Swedish a.i. was problematic not simply in terms of politics or scientific development, but also in terms of its practical usage. Many bulls failed to use the artificial vagina as intended, and their refusals to be 'fooled', their 'sluggishness', their 'suspicion', their 'weak thrusts', their 'sewing machine' performances, and many other recognizable behavioral mismatches with the technology were formidable obstacles to its deployment. Far from a frictionless invention, failures defined much of Swedish a.i'.s early trajectory; failures that also highlight how the use of the artificial vagina in a.i. made bulls' active and reliable participation in breeding more important than before.
Second, the veterinary expertise that was mobilized to overcome these obstacles did not do so by trying to manipulate otherwise-desirable bulls into breeding, nor by attempting to develop technological fixes that would bypass bulls' inability or resistance. Concerned with the possible inheritance of infertility, they instead sought to redefine a desirable bull as one that, whatever its other qualities, also had suitable libido and sexual behavior as evaluated using the artificial vagina. In Sweden, therefore, a specialized veterinary expertise centered on the sexual behavior and agency of a.i. bulls developed, and such behavior and agency came to be seen as crucial to the developing a.i. system. Without denying the fateful hierarchies intrinsic to farming, the actual human-animal encounters that constituted semen collection therefore came to build on a certain respect for the bulls' autonomy.
Third and finally, the veterinary redefinition of what a desirable bull was, and their centering of bull agency, helped undermine established institutions and structures of cattle breeding. This is a broader point, speaking to the complex relationship between science, technology, production, and reproduction in modern farming. Reproduction, as I noted (with Gabriel Rosenberg), is at the heart of agriculture. In dairy farming, its effective functioning is a direct determinant of productivity and thus of profitability. This helps explain why the veterinary arguments that linked bulls' failures to use the artificial vagina to particular practices of pedigree breeders were more than just technical points. Bringing out the examples of sluggish and unwilling bulls as evidence that breeders' careless practices interfered with reproductive functioning amounted to a reckoning with the entire system of exterior-focused breeding. Actors seeking to promote a more quantitatively oriented breeding could seize on it to attack the institutions and influence of more traditionally minded breeders. This helped them bring about a rapid shift. By the end of the 1950s, the exterior-focused pedigree breeders had lost most of their influence, being replaced by those in charge of a new system that built on progeny testing and the processing of increasing amounts of quantitative data.
Taken together, I argue that these findings provide a new perspective on the history of artificial insemination. They show that the particular forms of human-animal interaction that the technology entailed, and problems related to such interactions, could have strong influence over how a.i. developed as a mode of breeding. To put it more directly: had most bulls served impeccably in the artificial vagina from the outset, the history of a.i. and cattle breeding in Sweden would had looked very different. The fact that they did not largely defined the development of both the technology and its institutional context before 1955. The extent to which the findings are relevant outside of Sweden remains, however, unclear. They deal with issues not addressed in earlier research. This might reflect that the problems with early a.i. were more prevalent in Sweden than in most other places, or it might -as Rosenberg has hinted at -reflect a source-related bias originating with the fact that animal breeders have generally been more interested in writing about their selection of animals than about their orchestration of actual mating. We do know that bull evaluations for a.i. that were somewhat similar to the Swedish ones took place elsewhere, too, suggesting that the problem of finding bulls suitable for a.i. was seen as serious enough to warrant a special procedure in many places. 69 The nature of the problem was, however, probably framed differently in different countries and some were more open to experiment with alternatives to the artificial vagina when encountering reluctant bulls. At the very least, however, this still supports my general point: that a. i'.s separation between service and impregnation actually made the regular supply of semen more important, meaning that the ability to serve became problematized in new ways. Ultimately, more research is needed about the prevalence of problems of use in early a.i., how these problems were interpreted and responded to, and what effects they produced in different national contexts. My study suggests that such problems could be of formative importance and that they deserve to be considered in future research.
As important as it is to study the development and use of early a.i., this is, however, only the beginning of the story. As historian of technology David Edgerton reminds us, histories of innovation are too often made to stand for the history of technology as such, producing, ultimately, misrepresentations of the relations between technology and society. The STS interest in users of technology tend, he notes, to fall into the same trap, limiting its interest in users to their influence on the development of new technologies. In contrast, Edgerton stresses the importance of studying not just users' influence on innovation but 'technology-in-use' over the longer term. 70 The artificial vagina happens to be a good example of technology-in-use, since it still -largely unchanged from the 1940s -undergirds artificial insemination today, and indeed has become vastly more important as there is practically no natural breeding of dairy cattle anymore. But if gathering semen looks similar, the practices and economics of breeding have changed tremendously. Deep-freezing of semen, introduced in the 1950s and 1960s, has removed all temporal and spatial constraints, even allowing for the use of semen from bulls no longer alive. These days, farmers in Sweden can even order their frozen semen doses from a handy web shop. 71 As for breeding itself, biotechnological and genetic developments have opened a slew of new possibilities for controlling its trajectories. As such, studying the introduction of the artificial vagina and revising our knowledge of how it opened the door for transforming breeding is a necessary first step, but only a first. What follows is the larger and ultimately more important story of the immense transformation of breeding and of cattle lives that followed as the artificial vagina went from technologyin-development to technology-in-use. It remains to be adequately told.