Complementary Innovations And Generative Relationships: An Ethnographic Study

The paper presents an in-depth investigation on a promising innovation, kervit, in the histolical context of the booming ceramic industrial district of SassuoloScandiano (Emilia-Romagna region in Italy). The kervit was introduced and patented by a brilliant inventor operating within, and then leading, one of the first cdc companies of the booming district The innovation had strong technological advantages aqd good market potentialities, but in the middle of the 1960s, right in the stage of sharp growth of the ceramic district, it was unable to exploit its market success, nor was the inventor's company able to survive. The joint use of the ebographic mathod and the notion of generative relationships, put forward by Lane and Maxfield (1997), marks an original contribution in the analysis both of the emergence of learning processes within a local productive system and of the role of dynamic complemwtarities in fostering the innovation dynamics.


INTRODUCTION
Why is it that, after two decades of successful use on an industrial scale, with a couple of dozen licences granted to h s in various overseas countries and a marked degree of competitivity in terms of cost, a technology is abandoned even though several of its basic characteristics will become standard in subsequent decades? What are the conditions in which patent protection of a production method can actually bring about a decline in the use of the patented technology?
T k s paper offers an answer these questions by pointing out the irrelevance of natural technological trajectories in the process of innovation and highlights the way in which the technical developments under discussion present marked discontinuities. These can be explained by the fact that in the development of a new technology there are not just elements of technical necessity but, rather, path-dependent processes come into play influenced by heterogeneous and contingent factorstechnical, social, economic factors and, to some extent, also political and institutional factorsthat interact to produce the particular form of technological development which we finally observe.
The process whereby knowledge is created and transmitted is a critical aspect in the analysis of how innovation comes about. The dynamic dimension of this phenomenon requires us to identify the analytical tools which will enable us to define and interpret the historic sequence of different, but endogenous, spatial interactionsbetween agents and between agents and artifactsby means of which codified knowledge, tacit knowledge, technical practices and cultural values are interwoven and fused into a specific local fabric which sustains and fuels innovation. To understand the laws of the aggregate dynamics of localized technological change, in this paper I shall adopt an ethnographic method and shall use the notion of generative relationships put forward by Lane and Maxfield (1997). This notion provides an appropriate language to single out and describe the elements that define the dynamics of changes of the agents and artifacts space.
In section 2 the paper presents an overall picture of the topics to be discussed in the empirical analysis; section 3 illustrates the relevant features of the ethnographical method I have used in collecting the data for this study and summarizes the Lane and Maxfield's notion of generative relationships. Focusing on changes in the agent artifact space in which kervit emerged, section 4 outlines the social, technical and economic features of the k e~t technology, while section 5 points out the critical aspects marking its demise. Lastly, referring both to the d y d c s of innovation in a local production system i d to the more general analytical aspects of the innovation process, section 6 summarizes the chief conclusions to be drawn from this study with regard to three main issues: the need for continuous monitoring and fostering of generativeness of relationships, the , emergence of new competences, the social .dimension of patents and dynamic complementarities.

AND DYNAMIC COMPLEMENTARITIES
The empirical research concerns the case history of kervit* , a radical innovation in the production of ceramic tile which was first tested by an Italian firm at the end of the 1930s and protected by various patents obtained at the end of the 1940s. For about twenty years, the firm that had initially patented and introduced it (S. A. Industria Ceramica Veggia) was also responsible for building all the machines and equipment needed to improve the technical and economic efficiency of kervit. After two decades of improvement in its use on an industrial scale during which a couple of dozen licences were issued to various firms in European and Latin American countries, the kervit technology went out of use owing to the closure of the company that had patented it.
When k e~t was abandoned in the mid-196Os, Italian tile production was being launched on the rapid growth that has characterized one of Italian industry's most dynamic districts -.namely, the ceramics district of Sassuolo-Scandiano (in the provinces of Modena and Reggio Emilia in Emilia-Romagna region). The history of kervit, therefore, takes us up to the opening stages of a local production systemt and enables us to see how the process of training technical expertise within the firm that introduced the innovation gave origin to wide spillovers that allowed the sedimentation of massive tacit knowledge in the local production system. In particular, ke~t'became a shining example, not only of what to do in technical terms, but also of what not to do in economic and social terms.
Let us fist of all consider the overall framework in which to locate the kervit story: the birth of the ceramics district, a district strongly oriented towards exports and where interaction between producers and users of machines have played a central part in the development of the system (Russo, 1985). In the initial stage of development of the ceramics district, * The name k e~t is an acronym composed of the first thne letters of the Greek word keramos (ceramic) and the Latin word vitnun (glass).
the main worldwide producers of ceramics machines were large German and British engineering firms. Wlth the expansion of the market, concen-@at& in a very limited area, these firms were joinedand shortly replaced*by new firms born in Italy in the ceramics district in order to produce a vast range of machines and equipment devised ad hoc, with the aim of meeting the growing demand for automation in the tileproducing process, but also of extending the range of praducts, reducing energy consumption and satisfying the ever-increasing environmental regulationst. Even now, these engineering firms, mostly small*, generally specialize in developing and setting up one or more stages needed to produce a single type of machine, and rely on a complex network of engineering subcontractors also producing for other engineering industries in the regione.g.
automobile engineering for sports cars and l u x q models (produced by Femari, Bugatti, Maserati), tractors, machines for the food industry. The subcontractors for firms producing ceramics machines seldom do additional work for other sectors: the mechanical stages and components they produce generally occupy a lower band in terms of the tolerances requested. However, their proximity to engineering h s that are more exacting from the technical point of view represents a continual source of enrichment of technical skills, owing to the mobility of workers (Brusco, 1982;Russo, 1985;Bellandi, 1989). * Genm and British ceramics machine producers abandoned that specialization in favour of others in the machine sector.
t In a comparative analysis regarding Italy, Spain. France and Germany, the effects of environme'ntal reaulation on the comwtitiveness of the firms in the ceramic district are discussed by Russo it al. (1998). $When k e~t was invented there was no small business system; there wae only thee firms. It was not until the early 1960s that a host of new tile producers and ceramics machinery manufactwen made their appearance, spedalizing in production and forming a tight network of sub-supplier relationships. The reference to small business should therefore be related to the phase m which kenit development reached its peak As far as the size of the business is concerned, it should be referred to the size of the market. In the absence of economies of scale of a different nature, the years 1930-1950 saw technical economies of scale in tile production which were influenced by the capacity of the kilns, which at that time was around 1000 square metres a day for firing pressed material and around 300 square metres a day for glazed material. It was hence a very small dimension compared with the size of the market and not prohibitive compared with the investment necessary to construct a plant of minimal efficiency dimensions. When the 1960s witnessed a h t i c increase in domestic demand, the absence of bamers meant that a lot of firms set up in tke ceramics industry. F m the end of 1970s. the size distribution of tile-producing firms was modified by continual mergers and take-overs, so that today it is largely characterised by the presence of some large groups (Russo 1996). In characterizing the Sassuolo ceramics district it is thus necessary to keep in mind not only the multiple interactions between the engineering firins producing ceramics equipment and between these and the ceramics firms that use them, but also the interactions between all these and other firmsoperating in the vertically integrated sector of tile productionthat specialize in graphics, or transport services or in the many other commercial, administrative, technical and financial services. The majority of these firms are located within the ceramics district, which comprises an area of about 50 sq.km.
The extraordinary complexity of the system we see today could hardly be conceived in the 1950s, when the kervit technology underwent its maximum expansion and Industria Ceramica Veggia produced in-house not only all the intermediate goods needed to manufacture kervit tiles, but also the machines and equipment employed in the production process.
In tracing the history of kervit I shall show that the decision to abandon this production 'method was not due to the assumption that it had been economically and technically outgrown: the available documentation shows that it was remarkably competitive already in the 1950s (Vecchi, 1952) and recent estimates still assess it as a competitive technology*. The technology to produce kervit tiles contained some of the basic ideas that feature in the technical development of what, over the last twenty years, has become the leading technology at world level in the production of ceramic tiles: single-firing production. However, k e~t was not abandoned owing to its being ' *ay ahead of its time". My explanation is that kervit was dropped as a result of the gradual reduction, in the 1960s, of the generative potential of the relationships. that, in the previous decade, revolved around the kervit technology: the mental closure of kervit's inventor with respect to the space in which he operated led him to underestimate the entity of the ongoing changes.
To support my argument I shall propose a historical analysis of the social and economic context in which the k e~t technology emerged, developed and was subsequently discarded: this will shed light on the main actors and artifacts in the innovative process and will underline how interactions * The hypothesis in this cost calculation is a production of k d t tiles using present-day firing, glazing and movement techniques. in these conditions, the cost of kervit tiles would be competitive with' that of tiles of similar porosity and thickness (Cf. comment by engineer Franco Camevali in the second meeting with Dal Borgo, coded as 2DB. before 2DB47; see section 3.1).
between actors change in time the environment in which they operate. It will further point out changes in attribution on the part of agents both of other agents and vis-214s the artifacts, and will highlight the endogenous and exogenous conditions that alter the generative potential of relationships, marking the demise of the innovation in question. It is by now widely recognized that the historical analysis of technologies that are abandoned demonstrates elements which are generally less visible when success stories are studied*. Here, the case study of a failure enables us to ascertain to what extent the lack of continuous monitoring of the generative potential of relationships can influence the dynamics of the innovative process.
In this paper, analysis of the generativeness of relationships becomes also the key element for discussing the role of patents in assessing innovative activity in local production systems. The empirical analysis presented in this paper highlights that, even if there may be products and production methods whose original features make them suitable for patent registration, the full production and commercial exploitation of many patented innovations requires the construction of a network of relationships which are not only technical, productive and organizational, but also social and economic. Analysis of the social context in which the patented innovation has meaning for those who use itan aspect generally overlooked in the economic analysis of patent registrationsthus becomes a crucial issue in the researcht because such an analysis may actually turn out to be more conclusive than merely calculating the number of patents.
Economists by now tend to consider some of these aspects when observing that complementary assets or capabilities must be utilized in conjunction with innovation fleece, 1986; Antonelli, 1999): t~ take full advantage of the patent, the innovative agent has to decide whether to vertically integrate them with the innovation activity of the b, or to implement partnerships with the owners of those complementarities fleece, 1986). According to this view, these complementarities rare considered as already existing. In this paper, instead, I will argue that they emerge as an outcome of the interplay between the industrial dynamics and the dynamics of inno-vation. The basic idea is that a new product or a new production process nearly always needs to be accompanied by complementary innovations whi'ch are only developed if there is sufficient demand to sustain the effort needed to innovate. In these conditions, if the innovating f i r m patents the innovative product or production process, but does not generate sufficient demand to induce other firms to develop necessary complementary innovations, and moreover lacks the expertise to develop them in house itself, then patent protection actually restricts expansion right the way through the innovative process.
One reason for this restriction is explained in terms of network externalities, generally associated with networks of consumer goods users (Katz andShapiro, 1985, andArthur, 1989). But what the kervit story enables us to highlight is that such complementarities are also generated in the expansion stage of a new industry or new technology, when the increasing . demand provokes the emergence of new actors specializing in one or more stages of the entire process of production of machinery, intermediate goods, semihished products, and services needed for the new technology (Bonifati, 1999). What we have, then, is that complementarities of produc--tion are associated with the growing roundaboutness of the production processes (Young, 1928). In these conditions, if the patenting iim does not represent a sufficient source of demand, the patent protection may prevent the patenting f i r m from exchanging information with other agents who might otherwise contribute to solving technical problems which normally arise while the innovation is being fine-tuned (Rosenberg, 1982).
In this framework, we are interested in examining the complementarities emerging within the innovation process (Rosenberg, 1996). and not only those already existing and that agents are ready to use; in other words, we are interested in the dynamic dimension of complementarities. The kervit story thus helps us also to understand in which particular ways the lirms operating in a local productive system appropriate the benefits deriving from innovationways that require us to interpret the recourse to patenting in an ampler perspective than the one traditionally adopted by economists.

CONCEPTUAL FRAMEWORK AND METHODOLOGY
By abandoning the traditional methods of analysis we are forced to define the analytical tools and also the most appropriate vocabulary to be adopted in the research project. In particular, we need to take into consideration not only the agents' characteristics but also the history of interactions between those agents: in this case study it is thus necessauy to reconstruct the social, technical and economic processes within which the dynamics of change in the agent and artifact space brought about by the kervit innovation can be described and interpreted.
In collecting data relevant to the analysis of the kewit innovation I have favoured a method put forward by ethnographers, who, making use of open interviews, describe the subjects and the artifacts in the social and economic environment in which they operate, without imposing on the subject the conceptual categories of the person canying out the study*. The aim is, then, to "view the world through the eyes of the inte~ewees". In fact, if our objective is to understand innovation as a combination of changes both on the cognitive and on the smctural level, ethnographic analysis can help in defining an enquiry method ts understand how the actors in the innovation process perceive and categorize reality. Such a method enables us to perform a historical analysis of the interactions which constitute the relationships between the agents who initiated the innovation process.
In this section, together with the ethnographic method adopted in the research (section 3.1), the particular vocabulary that I use here (summarized in section 3.2) will be made explicit.
* One helpful reference for setting up this enquiry method is the work done by Spradley (1979) who outlines a methodology for ethnographic r e s w h : it ranges from how to decide on who to interview, to specifyinithe various phases of data collectio~ and the elaboration of relevant information. For a discussion of the use of the ethnographic method in analyzing situated actiom (in which the context the action develops in is modified by that action) see Suchman (1987). A wide ovemew on the characteristics of the ethnographic method is presented by the challenging first c h a m of the second edition of Agar's Professional S m g e r (1996).

Ethnographic method to describe and interpret the dynamics of innovation
Ethnography is not merely description, but also analysis and interpretation, this is why any study intending to call itself "ethnographic" must satisfy the two following conditions: a) An ethnographic analysis must utilize different types of data from several sources so as to have a "massive over-determination of pattern", without which it would be impossible to construct and interconnect the multiple %framesw (understood as blknowledge structures") needed for the analysis and interpretation of the phenomena of study. b) "New concepts have to exist at the end of the study that didn't exist in the original research problem": no abduction no ethnography, A@ reminds us @. 39).
Keeping in mind these two conditions, let us start by noting that the ethnographic method' entails using first-hand sources, study of which involves entering into relationship with whom is being studied, participating in what they do, and observing what happens. Obviously this has not been possible in the present paper. However, in reconstructing the events that took place between the 1920s and the 1960s, I have been able to draw on the fkt-hand experience of Antonino Dal Borgo, the kervit inventor, . whose readiness to co-oprate enabled us to compile a large part of the documentation needed for this case study. Around 8 hours of conversation were recorded during four different sittings. Following a preliminary literal transcription of the first two recorded interviews, the texts were reviewed and corrected by me, and subsequently revised by the interviewee and discussed in the two later sittings. These revisions were necessary in order to give greater definition to themes developed in later questioning. But they were also necessary in order to verify that in the literal interview transcriptions the punctuation had not altered the sense of the written text in respect of the spoken original. These revisions are given as annotations to complete the transcriptions of the first two interviews*. The preliminary texts are an important part of the ethnographic work, as regards both method and ethical level. Processing the material collected in the interviews brought out some unexpected elements (what ethnographers call "rich points") which improved our understanding of the context in which the kervit story unfolded and made it necessary to alter the working hypotheses. As well as Dal Borgo, it was also possible to have long interviews with Silvio Lusuardi, former mechanic at Cerarnica Veggia since the 19505, and Franco Carnevali, one of the first engineers to be hired by a ceramics firm in the Sassuolo district. After some years as consultant in an important office in Turin, Carnevali was called to the f i r m at Veggia by his brother-in-law (a son of Dal Borgo) in order to lend a hand in "tidying up" the father's technical domain. The experience acquired by Carnevali in the subsequent decades as production manager in one of the main groups of the district makes him peculiarly able to assess the elements of continuity and discontinuity that gradually emerge in the k e~t story. As well as the aforesaid material and interviews with various technicians, attention has also focused on interviews, performed in a previous mearch into innovation*, where fleeting reference was made to kervit. Those very elements collected in previous interviews aroused my curiosity on a particular aspect of the growth of the district: kervit was cited as a technology that comprised in a nutshell essential elements of what became the currently dominant technology, but nobody gave a satisfactory explanation as to why k e~t was abandoned. Here was a golden ogportunity for me to investigate directly Dal Borgo's experience: the available technical documentation would have been quite inadequate, being very fragmentary and imprecise; suffice it to recall that even the patents regarding kervit cannot befound, for kervit has been overtaken by so many other matters in Dal Borgo's life that he has lost all trace of these patents. merely his technical and economic decisions but also the social and, more generally speaking, the relational dimensions in which he was immersed and in which those decisions were taken: all this is described by Dal Borgo with the detachment of one who has lucidly reflected on those events over the subsequent forty years during which he took a personal part in several of the changes occurring in the world in which he operated. The fact that Dal Borgo has re-elaborated his interpretation of kervit has necessitated a delicate work of comparison of the various available sources of information to delineate the cultural context of the interviewees as they themselves perceive it, through. time.
Along this line of research, in the interviews it has been useful to approach several themes via "descriptive", "structural" and cccontrasb, questions. The udescriptive~ questions ask the interviewees to describe people, artifacts, situations and experiences which they have had directly or observed. Such questions are helpful to our research in contextualizing the inte~ewees' personal and professional history inside the firm where they have worked. In these descriptions reference is made to people and artifacts for which it is necessary to specify the attributions assigned them by the inte~ewees. The description requires temporal and spatial rigour (when, in what order, where). The "structural" questions are more technical than those above and require the interviewees to specify technical and conceptual aspects which enable us to understand the meaning of the terms they use. In general, the structural questions are designed to focus on technical and economic aspects of the artifacts discussed, but also on aspects of the relationships between agents and artifacts which are mentioned in the descriptive part. These questions help to define the identity of the agents and the "athibutions" that they assign to themselves, to other agents, and to the artifacts that populate their technical, economic and social space. The "contrast" questions are designed to highlight possible differences in the use of certain conceptual categories or expressions: they help to clarify the meaning of the terms used. That is the meaning of both technical terms, but also of the picture that emerges from the way the interviewee represents the agent and artifact space.
The history of kervit and its inventor, Antonino Dal Borgo, will provide the narrative device to outline how innovation comes about in a local production system, but to understand the k e~t story we must first abandon the idea of the inventor who solves each and every problem; he is, rather, somebody who formulates and solves problems which are compatible with his expertise and ideas*. This change of perspective is the key to understanding how it is precisely the combination of Dal Borgo's technical expkrtise and his ideas of the world which determined, both for better and for worse, kervit's development and demise. As soon as we are ready to unfold such a narrative, we are compelled to find an appropriate language to describe the many interacting elements involved in the dynamics of changes with regard to agents, artifact and to the context in which they operate: the narrative that follows w i l l draw on the vocabulary proposed by Lane and Maxfield (1997) and I shall now summarize their contribution focusing my .presentation on the notion of generative relationships.
33 Generative relationships and innovation: looking for an adequate language First of all we need to consider that any relationship between agents is made up of a multiplicity of interactions occurring at many levels (for example, personal as well as professional) and via many channels. Lane and Maxfield (1997) define as "generative" those relationships which can induce changes in the way in which those who participate in the relationship see their world and act within it, bringing about innovations which are generally characterized as new entitiessuch as, for example, new agents or new artifacts or new institutions. In these tenns, the analysis put forward by Lane and Maxfield is consistent with the definition of innovation suggested by Schumpeter (1934), with the vision of space of action put forward by Perrow (1961) and with the notion of technological systems proposed by Hughes (1983)~. Lane and Maxfield's original contribution lies in their specifying what constitutes a "generative relationship" and in which ways such.re1ationships can be created.
To outline these elements we must first b t d u c e a major aspect of the analysis proposed by Lane and Maxfieldnamely, the importance of the intelptetation of the significance that agents ascribe to themselves, to other agents and to artifacts: what Lane and Maxfield call "attributions". The world in which agents operate is defined precisely in terms of their perception of the context in which they act Within such a world the interactions between agents and between agents and artifacts define the structure of their space of action. Structural change in the agentlartifact space is  thus mediated by new attributions as to the identity of the agents and the meaning of the artifacts. The identity of an agent is defined by what he does (his function), how he does it and with whom, and for whom (his character). The significance that agents ascribe to themselves, to their products, to their competitors and clients and to all the other actors present in their world determines the possible space in which they act and the way in which they act. The functions performed by the agent define the "zone" of a space whose structure the agent attempts to change. The character of an agent can be specified by the means (agents and artifacts) that the agent mobilizes to achieve the transformation that he is seeking, by the means and non-material resources with which the agent influences other agents and artifacts to achieve the desired mobilization.
The significance an agent gives to an artifact regards, firstly, the use that agents make of the artifact+. The identity of an artifact is thus defined'by its use, by who uses it and for what purpose together with (or instead of) which other artifacts. For the person designing or producing the artifact its function alone is not a sufficient attribute: he must also consider the way in which the artifact relates to other artifacts which comprise it. This has also to do attribution of functionality.
Generative relationships are the result of interactions between agents: entire companies, but also between departments or individuals inside or outside those companies. These multi-level interactions between agents and artifacts (both inside and outside the company) do not necessarily mean that the result w i l l be the creation of a generative relationship or its maintenance: it may be that within a firm there are departments which in their interactions with other agents and artifacts promote the formation of generative relationships while other departments stifle that formation. This is why the changes which result from generative relationships cannot be predicted on the basis of the type of lcnowlehge possessed by the agents involved in the relationship. Such changes are in fact the result of a process in which the technical, economic, social and institutional dimension are components which do not operate independently of one another. The interpretation of the result of this process thus requires a knowledge of the structure and the history of the interactions which make up the relationships between those agentst: since we need a description of the * This clarifies the sense in which an existing product can be consided an innovation, as Schumpeter, for example, asserted.
'dynamics of interactions, the ethnographic method w i l l provide the most suitable way of acquiring that knowledge.
%at makes generative relationships important for our analysis is that they induce changes in attributions, and these changes, which are frequently of a cumulative nature, in turn create conditions for new generative relationships. This boot-strap dynamics is a major feature therefore of the structural change that takes place in agent md h f a c t space. In order to assess which relationships have generative potential, Lane and Maxfield identify five preconditions.
1. For the tiansfoxmation of some particular zone of the space, those involved in the relationship must share, in their activities, a focus on some artifact or agent (aligned directedness); 2. The relationship must combine differences between the agents in terms of expertise, attributions or access to particular agents or artifacts (heterogeneity of agents). This can help to generate new expertise or attributions as a result of the relationship. 3. Agents must seek to develop a recurrent pattern of interactions from which a relationship can emerge (mutual dinxtealness). Their willingness to do so depends on the attributions that each assigns to the identity of the other. In this context, mutual trust helps but is not a precondition. Actually, it may be the result of the interactions through which agents realize that they can derive benefits from the relationship that is being generated. 4. It is necessarily for those involved in the relationship to have discursive relationships (permissions). This must also happen outside the conventional exchanges which are generally confined to requests, orders, declarations. This condition is fostered by a company structure which envisages a distributed control. 5. Discussing matters of common interest can prove more incisive if the agents have the chance to interact in at least one activity which sees them working together (opportunity for common action).
Lane and Maxfield stress the fact that these conditions must be constantly monitored because the agents must be in a position to interpret the changes that are the direct result of those relationships. Monitoring can suggest ways of nurturing and maintaining generative potential of relationships, because as soon as an agent discovers that changes are occurring, over a period of time, in respect of attributions (assigned to himself and to other.agents and artifacts), he will also try to identify the source of these changes in the various relationships he is involved in and may also discover a way of feeding those relationships which will possibly give rise, in turn, to further changes.
The notion of generative relationships was introduced by Lane and Maxfield to analyze complex situations, where continual and rapid changes take place in agent and artifact space. In this study, I shall make use of that notion because it provides an adequate language to describe the dynamics of innovation. In particular, I propose to investigate the five conditions which are a feature of generativeness in relationships in order to assess their efficiency in creating changes in agent and artifact attributes, in new relationships with other agents, and in common action to create changes in the agent and artifact space. This approach to the k e~t story will enable us to recognize how it first emerged and why it came to grief.

RELATIONSHIPS
Let us focus now 'on the main aspects characterizing the dynamics of changes in the agent and artifact space in which kervit emerged, In particular, kf2~t.S inventor, Antonino Dal B o w , and the interplay of changes in his attribution to other agents and artifacts will be presented in section 4.1, highlighting the technical, economic and social characteristics .of kervit; section 4.2 then investigates the elements of heterogeneity, aligned and mutual directedness, right permissions and opportunity for common action that were crucial for the generativeness of the relationships -between technical genius Dal Borgo and Maurizio Korach, an intellectual with a wide range of relationships and a leading figure in chemistryup to the peak of the many licences for kervit issued to dozens of overseas companies in the early 1960s.

Technical, economic and social characteristics of kervit innovation
When in January 1928 Antonino Dal Borgo arrived at Sant' Antoninoan unknown backwater on the left bank of the river Secchia, in the Italian province of Reggio Emiliathe Industria Ceramica Veggia S.A. had been operational for no more than four years. With little less than one hundred employees, it was flanked by two other tile factories, originally part of the Rubbiani k n , which had been in business for more than two centuries. Hie was called in by Ceramica Veggia to replace the factory foremana y o h g chemist who, like himself, had studied at the Scuola d'Arte Ceramica in Faenzano doubt qualified from the technical point of view, but not up to the task of co-ord&ting and supervising the workforce. After only six months, the twenty-two year old was already making a name for himself when he introduced his first ceramic production innovation by substituting the age-old tradition tin glazes, then become expensive, with arsenic glazes. Just after graduation, Dal Borgo had gained work experience in a company in Femra and it was probably here h i he learned that a chemist from Forli had med to use arsenic instead of tin. Following a brief series of experiments conducted in his small laboratory at Veggia, Dal Borgo began producing arsenic glazes, obtaining from the frit a white opaque product at a cost considerably lower than that of the tin glaze*, and this at a time when the glaze was the principle element of production cost. When in the Thirties he used zircon instead of arsenic, he created the glaze that became hown as "Sassuolo white": a glaze, used for many years in Italian tile production, which made it possible to obtain, at low cost, a tile with'a white surface applied to the red tiles made of the local clayst.
From Dal Borgo's description of the way in which he discovered these new glazes it emerged that in some cases he simply applied existing techniques in new fields (the properties of arsenic or zirconium), in other cases he found solutions by breaking down the problem into its chemical components* and by exploiting the chemical and physical properties of the various compounds. The experimental conditions offered by the small chemistry laboratory he had at the factory were adequate; he availed himself of an assistant's help andusing rudimentary equipment for experimenting and measuringhe did everything himself. He was very young, and the expertise acquired at the Faenza school of ceramic technology could be used to experiment directly and mate new things ¶ * "In the molten state, the frit spontaneously forms white crystals (lead menate) in the glass, thus -lacing the tin oxide, which acted as a covering, and costing less. The idea was prompted by the fact that tin oxide was very expensive and difficult to get hold of." [1DB20]; "uulike arsenic, tin has always been a rare and costly commodity, and at that time tin oxide was used only in cold glazing." [1DB21]. This inno\iative spirit is the hallmark of everything Dal Borgo did*. However, his innovative approach significantly diverged from that of many other c h e h t s who, in the 1960s and in the 1970s, were to enter ceramics companies in Sassuolo and who developed new colours and surface glazing effects, at times drawing inspiration h m production defects in order to obtain original decorative effectst. In Dal Borgo's case, he developed not so much a surface glazing effect, as a new production methodcalled "vitral"*by means of which the tile, instead of being a biscuit body with a vied surface, becomes simply the viQ%ed layer only a few millimetres thick In the case of fist vitral, and later kervit, Dal Borgo's idea stemmed from the combination of two spheres regarding, respectively, the production and the use of the ceramic wall tiles. With regard to the production sphere, Dal Borgo had to face what, at that time, was a common defect in tile production: the "scaling" which formed when the vitdied part came away fiom the biscuit body. The defective tiles were wasted and this increased production cost This defect was due to the different dilatation coefficients of the glaze and the biscuit body it was applied to. With regard to the use of the tiles, Dal Borgo observed that, once fixed on a wall, the ceramic tile presented a smooth, easily washable and therefore hygienic, as well as aesthetically pleasing, surface. By focusing only on what he interpreted as the most relevant functionality of the tile, and not so much on improvements on the traditional production technique, he then created the technical conditions which made it possible to achieve the scaling effect over the entire surface of the tile, and in a systematically controlled way, thus obtaining a tile that was little more than a vitrified layer%.
* He patented many of his inventions, the most recent, patented at over ninety years of age. was a hearing aid, whose prototype he is using to his own personal satisfaction. t Until the mid-1960s. the technical management of many ceramics firms was largely under the guidance of a chemist who was often the factory manager as well, widely thought of as a "hands on" figure, up to his elbows in glazes and newly invented paints. The practice of buying in glazes and paints from specialist producers is relatively recent All d c s firms produced glazes in house, and during a technical development phase in which a large p r o m o n of processes wge carried out manually, glaze production was the most technically complex part and required specific training acquired partly in vocational and technical schools and partly from working alongside the factory chemist as an apprentice or assistant.
$ This name was composed from the k t four letters of the Latin word for glass (vlmrm). The ending (al) added a touch of modernity coming, already in the 1930s. from the English-speaking world.
q It is observable in the history of many inventions that the intersection of two different planes marks a decisive step in the process which produces new entities. For a telling analysis of this process see Koestler (1964).
To obtain this technical condition, Dal Borgo concentrated on the properties of the glazes and of an anti-adhesive (imposed of 15% bentonite and 85% magn&site*) which was applied to the surface of a rektory plate whose surface area was that of the tile he wished to A layer of glaze was then p o d onto the anti-adhesive. Decorations were then applied, using stencils obtained by cutting up pieces of greaseproofed paper, as was the traditional practice in the production of majolica tiles. Only at the end of firing did the &-adhesive react, causing the vitriiied d a c e to come away 'hm the refractmy plate, which was then used again in further production.
This technique to produce vitral, which Dsal Borgo patented in the spring of 1935, was thus based on a radical change in the molding procedure, using pouring rather than pressing. Pouring is, incidentally, a well-known technique in ceramicsone need only recall that sanitary furniture is produced in this way. The difference is that in the vitrd technique the mold used in the pouring stage was no more than a refkxtory plate. The pouring technique generally makes use of chalk molds for their capacity to absorb the humidity of the preparation poured into thtm, but chalk would not have been suitable for the vitral procedure since it would not have been usable during the firing phase. This is why Dal Borgo developed the system using a refractory cordierite plate which would be both porous (and hence able to absorb the humidity of the preparation poured onto it) and also resistant to the thermal shock involved in the use of several firing cyclesS. These plates, used in about 100 production cycles9, were pro-* Dal Borgo obtained this product by experimenting for about a ten day period. He knew the pmperties of the materials: a plastifier (bentonite), a refmcmy (magnesite, a dolomite easily located in the area since the stone is typical of the river Secchia). He knew he had to use only small quantities [lDBZS and note]. h i n g the 1930% scientific work on the characteristics of bentonite was very confused and imprecise (d. Vecchi. 1952, p. 19). Employment of this clay. then, was c e d y an element deviloped in an original way by DA BO&O.-t The anti-adhesiveapplied with a spray gun (a very thin layer was sufiicient)was absorbed by the refractory plate.
[1DB41] $ Coderite is a silicon aluminate of magnesium. In the early 1950s there was much animated scientific arrmment as to the emnlovment of coderite mixtures in industrv. Althoueh it had been trled G1taly in the mid-ni;e&th century, Vecchi (1952, pp. 21-2)kso = that in the 1930s Maurizio KO& and G. Fuschi & v e l d the use of cordierite mixtures in order to improve resistance to the sudden temperahue &ang= of electrical insulators. This use (covered by patents of 1932.1933 and 1937) was also put forward by Korach at various meetings and in a 1934 article in the journal Elemotecnica. Hence, in the 1930s. its only other industrial use seems to have been that made by Dal Borgo who, once again, anticipated the scientific debate that was to develop at the end of the 1940s. The link between Dal Borgo and Korach, of which we wiU speak later on, suggests that Dal Borgo knew about the research done by Korach and Fuschi, nonetheless Dal Borgo made original use of wrdiedte mixtures in the production of supporting plates used to produce vitrsl and, later, kervit tiles. g[ Cf. 1DB37. ducedinside the firmusing filter-presses for drying, friction presses for molding and small intermittent muffle kilns for firing. Thus a batch of plates was produced each time it was necessary to refurnish the stock depleted by use. The vitral idea was decidedly innovative and the patent added much to Dal Borgo's growing reputation as an inventor. However, the originality of the new technique met a serious setback when the materials were first positioned: the use of mortar-based products caused the tile to come away from the wall*; a problem which also arose in the case of other new products used in those years for cladding, for example Opaline, large, long opal glass plates 6-7 millimetres thick used as a wall covering in bathroomst. While cladding materials agents welcomed the vitral innovation, those who had to put it in place, i.e. the tile fixers, were not quite so keen.
Alongside the traditional technique of tile production, over the centuries a laying technique had been developed using materials suitable for fixing to the wall a tile consisting of a robust body that " s u p p o~ the glazed part. Dal Borgo thought the development of a material more suitable for fixing vitralS a secondary consideration, preferring instead to work on a technical solution which would modify the product itself. Side by side with the traditional production of majolica, Dal Borgo thus continued his experiments to solve the problems regarding vitral, and after more than ten years of experimentation he proposed a new ideak e~twhich kept faith with his original idea in terms of the functional features of the tile and in terms of molding by pouring, but he added a biscuit body for the vitrified part.
The kervit technology envisaged applying to the rehctory plate, in sequence: first, the anti-adhesive layer; second, a mixture of ceramic preparation some three millimetres thick (which would form the body, as it was called, after firing); third, a layer of engobe and, finally, a layer of glaze. At the end of the firing cycle the refractory plate came away and the k e~t tile had more body than vitral. It is worth noting that in ceramics technology the white clay referred to as engobe had been known since antiquity and was used to cover the coloured surface of bowls and drinking vessels. It was Dal Borgo who was the first to use engobe in the production of tiles since it enabled him to use ordinary materials for the body (as for example discarded tiles or glass fragments which came from Murano). Without the engobeto keep the glaze whiteit would have been necessary to use pure raw materials which were expensive2.
While molding by pouring was a key change i n the kervit innovation, the other innovative dimension in the kervit case-history is clearly the single firing of the biscuit support and glaze: one of the main ingredients of subsequent single-& technologyt. In the case of vitral and, later, that of kervit, Dal Borgo relied on firing techniques used for glazed majolica tiles which were .then fired at maximum temperatures of between 960 and 980 degrees centigrade. To achieve the required resistance in the body at these temperatures, Dal Borgo employed vit~eous based clay preparations*. However, he used a firing time of around two and a half hours, thus much shorter than that then used in firing majolica tiles: this was technically possible and led to a marked reduction in production costs. Reduced firing time was a considerable new departure from techniques then in use: in a tubular kiln ("a passo di pellegrino"), which was then the fastest, firing time was never less than eight hours, while in the tunnel kiln which was subsequently used, firing glazed tiles required mund sixteen hours.
In another kind of kiln, where saggars of glazed tiles were transported on sliding batts, firing time was initially around twelve hours and, though eventually reduced to five or six hours, were much longer than those used in firing by kervitq. In interviews, Dal Borgo stresses that firing time in majolica tile production was fixed by convention rather khan by technical demands. The change he introduced was not the result of specific experimentation but rather the result of what he had theretofore intended: he was using theoretical knowledge and experience which he had probably never even verbalized, but which were implicitly present in his mental apparatus. This knowledge emerged when he concentrated his attention on the specific firing technique to be used for a product which differed from traditional products. At this point be discove~d that a firing time of no more than two and a half hours was sufficient, where "discov-* Cf. 1DB36. The engobe was subsequently used in the single-fire process. t Cf. 1DB121. * Cf. lDB37. 9 Cf. 1DB63. Q Referring to how he obtained this result, Dal Borgo replies that "it's one of those things that just happens like Ute way wheat grows" [Cf. 1DB63 an8 note] precisely because, with hindsight, discoveries seem obvious. ery" means "revealing something that has always been there, but which was hidden from our eyes by the blinkers of habitw*.
The technical literature of the time describes the k e~t technology as technically and economically superior to the technique then in use: introduction of glass in the ceramic mixture provided better cohesion between mixture and glazes; the length of the entire line was only 170 metres as against the 250 of traditional majolica and 300 of terraglia production; the manufacturing process was simple and almost completely automatic. All of this enabled a lower cost as compared with majolica and terraglia (Vecchi, 1952, p. 22). As early as the 1950s, k e~t worked out economically competitive. Suffice it to mention that, since kervit production costs were lower than those of majolica tiles, Dal Borgo had decided to sell it at a price which was proportionately lower, aiming at competitiveness based on price which would at all events have covered the cost of the licence (about 5% of turnover) of licensed k e~t producers. All this is confirmed by our research, except for two elementsthe simplicity of the p m s and its level of automation, which were to be critical in the development of k e~t , but before discussing these elements in the following sections let us complete the setting of the stage by allowing Maurizio Korach to enter the scene.

3 A window on the outside world: Mauriao Koracb enters the scene
To develop the kervit procedure, Dal Borgo employed some of his sons as assistants, as well as resorting to the mechanical and woodworking skills of those who worked in the Veggia workshopst. This meant a nucleus of around 25 carpenters and mechanics who helped in constructing the machinery and equipment necessary for production.
To this in-house staff must be added the collaboration of Dal Borgo with Maurizio ~orach*, a leading figure in chemical engineering, but also inter-* Cf. Koestla (1964), p. 98. When in the early 1960s Dal Borgo worked as a consultant for the firm which had taken o v a Veggia, he used the experience he had gained with kervit by personally contributing to the production of the first single-fire wall tiles. But, though there are several key features which suggest continuity with kervit, the developments in this production technique, wbich from the 1980s on has become the most widespread in the district, are also linked to other innovative contexts. For an analysis of the technical and economic conditions in which single-firing developed see Russo (1996). t Cf. interview with Lusuardi. Korach (1964). products of the Hungarian factory at Zolnay, of which Wartha was consultant; at that time Waaha was considered the leading world expert in ceramic chemistry. This was therefore an important opportunity for Ballardini to involve Korach, Wartha's pupil ("the wizard of Budapest" as Korach calls Wartha) in the projects concerning the art school that had just been set up (cf. Korach, 1964, andBiavati, 1976). Korach (1964) recalls how the collaboration between himself and Ballatdiai was a happy encounter between the historical-artistic view of production, Ballardini's, and his own technical background. From that encounter, ma& possible by their common "humanistic ideal". arose the Experimental Chemistry Laboratory of the Scuola d'Arte Ceramics at Faenza that Korach directed for ten years and contributed to making a research centre at international level. $ Already when he worked in Faen& his dissent fro; F&m was well known enough for the local Fascists to make him drink castor oil at the entrance of the School. Later, when he was in Bologna, he was advised by a friend that fascists were plotting to kin him (d. 1DB95 and Biavati, 1976).

$The biographical information about M a d o Korach was collected during the interviews with Dal Borgo, and in the papers by Vecchi (1952). Biavati (1976). Polinszky (1976), Vecchi (1988). Autobiographical notes can be found in
cello Cora enabled him to continue with his work as consultant to Italian chemical firms until the race laws came into force in 1938, whereupon he decided to leave Italy. He returned to take part in the war of liberation, but was arrested and imprisoned at San V~ttore. After the liberation, he resumed his academic activity, but his possibilities in Italy were limited by his connections with support for the Communist Party and his scientific ambitions, so he agreed to cooperate with the Hungarian government in the creation of the Central Institute for Building Materials which, thanks also to his work, was for decades the leading research centre in Europe in this field.* Korach is important in the kervit story for at least two reasons: firstly with regard to patent activity, secondly concerning the use of the "a passo di pellegrino" kiln (to which we shall make only marginal reference). As he had done previously for the vitral process, Dal Borgo entirely entrusted to his master and fiiend Maurizio Korach the task of preparing the technical specifications to be included in the kervit patent registration documents. He described in detail the technical procedures he had refined during the long period of experimentation and Korach couched them in the formal language required for a patent application. Since he was heavily engaged in the running of the Veggia factory, Dal Borgo also delegated to Korach the task of publicizing the features of the patented kervit systemt. His academic position held before the War togetherwith his fame as brilliant research scientist' and intellectual with a wide network of international acquaintances opened up numerous opportunities for Korach to sell the kervit licence abroad. Korach, who was well connected in the university world and frequented writem, painters and intellectuals, was put into contact via Swiss friends with John E. Mackenzie, a Scottish nobleman who taught parapsychology in Geneva and who had a wide network of business relations at an international level. Mackenzie became a key figure in the distribution of the kervit patent in the 1950s. In addition to Switzerland, Germany, France and Great Britain, the kervit licence was also sold in Israel, Venezuela and Brazil. The sole licensee and producer in Italy was Ilsa of Albissola in the province of Savona. Even after his retum to Hungary, Korach remained in contact with Dal Borgo and for several years took part in the international meetings that were held annually to diffuse the improvements introduced by Dal Borgo to the k e~t licensees.

ITS DEMISE
The relational aspects are one of the keys to understanding what happened to kervit and the subsequent turns the history of Veggia was to take. The economic and cultural environment in which Dal Borgo operated since the 1920s was that of an Italian province considered a depressed area from the economic point of view. Agriculture was the chief economic activity and the few industrial companies that operated in the area astride the provinces of Modena and Reggio Emilia on the banks of the river Secchia were a handful of fruit conserve firms, the producer of Sassolino (a liqueur well known in; the area) and three other ceramics W s which had set up beside the companies who had been first to arrive in the area When Kervit was discarded, in the mid-1960s, the social and economic environment had markedly changed: a strong increase in demand of ceramic tiles and a relatively simple technology characterized the surging ceramic tile district that was offering everybody a chance to make handsome earnings from the transformation of clay into tiles. Why was kervit discarded just when the district was emerging with its incredible high rate of growth and profits?
By telling the story of Veggia's failure, section 5.1 introduces a first answer to this question: kervit was discarded inter alia because the fixm that had introduced it failed. We must then give account of the factors weakening the competitive position of the company headed by Dal Borgo, up to its failure. In this section the main focus will be on the description of the innovation dynamics in an environment that has changed because the innovation has initiated a series of changes and also because endogenous changes have affected the agents and amfact space here examined. Section 5.2 argues that the strong hierarchical organizational of Dal Borgo's company certainly contributed to its exclusion from external cross fertilization; section 5.3 presents the peculiar process of collective inventionemerging in the surging ceramic districtfrom which Dal Borgo kept himself and his company apart; and, finally, section 5.4 describes the way in which other agents, the press producers, were able to foster new generative relationships which contributed in displacing kervit space.

The failure of C e d c a Veggia
The long period of experimentation had allowed Dal Borgo to develop a procedure that could be used on an industrial scale and Dal Borgo intended to take advantage of the financial benefits which would result from his invention: in 1947 he patented the kervit process and obtained a new contract with Veggia in order to insert a clause which, apart from recognizing his right to benefit personally from the patent protection, also gave him a minority shareholding in the company. Nine years later Industria Ceramica Veggia dismissed Dal Borgo, accusing him of exploiting for personal gain a patent which, according to legislation then in force, should have belonged to the company of which he was an employee. He made counter claims against the Veggia majority shareholder, since the special clause inserted in the contract would have secured a verdict in his favour. However, the matter never came to court: without his vital technical guidance, within the space of several months the company was on the verge of bankruptcy and, seven months after being dismissed, the old proprietors withdrew the charges and Dal Borgo returned to Veggia in 1957 as Chairman and major shareholder*.
Dal Borgo retained that position until 1967, when the Modena court declared Ceramica Veggia banlaupt, a tunring-point in the kerist story.+ While involved in the technical management of the firm, Dal Borgo had engaged in illegal accounting practices, very widespread at that time. Official inspection revealed that a sum of more than 500 million lire was unac- order to give kervit's British licensees time to weigh up a takeover of Ceramica Veggia, but the request was turned down by the court. Whereupon Dal Borgowho had no doubts of the firm's technical solidityappealed and offered to repay the creditors 50% of what they were owed (and in the following year the firm's management did indeed recover an amount exceeding 100% of their loans). The appeal, too, was refused, but the bank~~ptcy, declared on 21 April 1967, was not paid off till nearly twenty years later, with a mere. 10-11% to the creditors. Following the bankruptcy, Ceramica Veggia still went on producing tiles for a year, but soon switched to manufacturing colours, without Dal Borgo. The bankruptcy was due mainly to bad management, but the economic and administrative elements alone are not enough to explain the vehemence shown by the receivers in this case. To provide a convincing explanation it would be necessary to adduce much fuller documentation relative to the bankruptcy proceedings, as well as conducting a historical analysis of the political and social context of those years. 'This all lies outside the area of this study. Here I want to point out an aspect of the Veggia closure that takes us into a context widely discussed in economic literature inspired by Schumpeter, where the economic figure of the innovator is seen as analytically distinct from that of the inventor and stress is laid on the importance of the economic aspects of innovaSive activity. Though authoritative and extremely clever, Dal Borgo found it very difficult to deal with bankruptcy; his technical capabilitieswhich for forty years had enabled him to manage and then to buy a flourishing firm in continual expansionsuddenly appeared insufficient*. While still remaining an irrepressible inventor, Dal Borgo failed to become the complete entrepreneur, precisely because he was unable to take account of the multiple changes (including the economic ones) that resulted from his inventions.
* Not that hints of the need to change tack were lacking. Suffice it to think that his sons, close collaborators in the management of the firm, were aware of the need for a management that should add economic consistency to the extraordinary techid capacity of the inventor. l%at is why they suggested he hire Gianfianco Camevali, a young engineer with good experience of industrial oqpnhtion m the metal engineering sector and stmngly concerned to remain working at Sassuolo. Camevali enthusiastically accepted the oppmmity to apply his methods of industrial organization to a completely virgin &tory; at that time, in the ceramics firms of the dimict thae were only three other engineers, all at M d . But his enthusiasm soon waned before the urnstant necessity to change the production programmes as they went up in smoke owing to the continual akations in the glazes suggested end imposed by Dal Borgo. To be sure, these alterations improved the produc~ but they made progmmhg of the whole production pnwzss, and thus fdfillment of the orders, extremely unstable (d. Camevali 25).

3 Enterprise organization models and social relations
An implication of bad management recalls our attention to Veggia's organizational model that, like other ceramics finns operational at the time, involved a strategy of centralized control. In particular, Dal Borgo centralized to himself all the technical decisions, together with the poor economic control: this gave latitude for opportunistic behaviour at the limits of legalityto the point where Dal Borgo himself was forced to intervene, dismissing 16 workers who had removed materials and equipment belonging to Ceramica Veggia that were subsequently used in other ceramics firms set up by its employees.* Aside fiom these workers, against whom incontestable evidence was brought, there had been other cases in the early 1960s of employees who had profited from Dal Borgo's negligence in day-today economic and administrative matters.
These cases of opportunistic behaviour signal a situation with no common vision shared by the workers of Veggia and markedly affected by the absence of right permissions and opportunities for common action. Although Veggia was for a long time an important centre of technical training for hundreds of workers, Dal Borgo had a hierarchical vision of the company's organization that allowed no room for initiatives h m any one but himself. His sonswho ran the organizational side of the companywere as rigid as their father in managing social relationships with the company's workers +.
Although, at that time, extremely hierarchical and centralized control was a common feature in many other companies, it is interesting to compare Veggia's organizational and relational model with the one prevailing at Marazzi, a ceramics firm that came into being in 1936 in Sassuolo, of size comparable to that of Veggia. Filippo Marazzi, the firm's owner, was an entrepreneur with previous commercial experience on a local scale. Marazzi made up for a lack of specific technical training by taking on young technicians and giving a chance to those with flair and zeal to have a go, tucked away in a comer of the factory, at inventing something of use for the industry. This was the case with Leo Morandi who, during the 1950s, thanks partly to the experience gained with Marazzi, rose fiom respected bicycle repair man to become one of the most prolific * Cf. 1DB 113 and Camevali 27+31 t Cf. interview with Lusuardi.
inventors of devices for glazing equipmenta. The Marazzi microcosm aspired to being also a model of social and religious integration: witness the building of an infant school for the children of company employees, the medcd clinic, and even a chapel where mass was celebratedt. But Pietro Marazzi (who had taken over from his father at the head of the company) also set up a research and development unit inside the factory which became, from the 1950s on, an important centre for the innovation process which subsequently spread throughout the district.
Unlike Veggia, Marazzi actively sought interchange with the outside world. That Veggia did not do so helps to account for the state of isolation in which Veggia found itself when bankruptcy supervened.

3 Small improvements within the firm vs. collective invention in the l d system
The fact hat Dal Borgo remained prisoner, so to speak, of his factory, was a choice initially imposed on him by the lack of opporhdties offered by the environment in which his work attitude had been shaped. In the interviews, Dal Borgo 'ustifies his choice as the answer to autarchy imposed by 1 the fascist regime ; autarchy was to become a cultural model vis-&vis the outside worldq, but in Dal Borgo's case it also became the organizational and economic ,model at individual company level. And it was inside the factory that a generation of technicians, mechanics and carpenters was trained, as well as foremen and department headss who subsequently became entrepreneurs or plant managers in the ceramics firms (or ceram-* But in subsequent decades many other firms were set up by former Marazzi employees, as for example the firm that produced engineer Mario Poppi's kilns, or System, whose owner Franco Stefani had produced the serigraphic machine at Marazzi and from there went into business on his own, becoming one of the most dynamic machinery producers in the ceramics industry. Cf. Russo (1996).
f It is as well to remark that, although there are no systematic studies of these events, from the information I have managed to collect, the patanalistic model adopted by Marazti seems no different from that of other large finns that were operating in North Italy, e.g. Falk at Sesto San Giovanni, Milan (cf. Bertwxlli 1997).
f Cf. 2DB7. Althougb concrete performance of the autarchy plan was modest, it was successful from an ideological and propaganda point of view, because many technicians and scientists interpreted autarchy as the manifestation of a plan for altering society based on the canons of scientific rationality. On this inkqmthon of autarchy see NIaiocchi (1998). 5 In tile production, all the operations were carried out manually and during the 1950s Veggia's size grew to as many as 600 workers. Although this fell to some 300 by 1966, Veggia's dimensions were those of a firm with a multi-level hiemchy in terms of the organhational control of work and factory discipline. ics machine firms) which in the course of the 1960s spriing up all over the area Being cloistered away inside the factory waspart and parcel of a @at innovative project, the vitral and kervit process .
Compared with the traditional production technique used for .majolica tiles, what distinguished kervit was essentially the fact that, apart from the molding/pouring procedure, which we shall discuss further below, it needed only one firing during which the thin layer known as body, the layer of engobe and the glaze were thus all.fired together. The firing cycle was considerably reduced (by about two and a half hours) in kilns which fired one layer of tiles at a timet. Stated briefly in these terms, kervit technology appears analogous with the rapid single-firing technology which was widely adopted in the ceramics district duripg the 1980s; and to some extent it is, even if marked differences and discontinuities are apparent in the transition from one technology to the other. The interpretation of these discontinuities once again requires that we consider the relational aspects of the case. * This project was to become a technical model so inbred as not to be recognizable as such even by the technicians who were formed within i t In the hundreds of interviews with technicians and experts on ceramic tile technologies, conducted by me over the last twenty-five years, I never noted a mention of kervit technology. It was first mentioned to me by engineer F m c o Camevali two years ago (and in a detailed way in the interview of October 1998). but not during the several discussions on the development of ceramic technology which we had had in the pmvious twenty years. I spoke of this technology with a degree student of mine, Guido Cattani, who was conducting interviews on the innovation process in the production of low' porosity single-firing tiles. We decided to collect the documentation that we gradually turned up on kervit. From the analysis of the transcriptions (6. Canani, 1996), I found four interviews in which M t was cited: the kervit story resurfaced in the memory of the technicians interviewed when they were refaring to a single-fire, low porosity prod-uct arriving on the market that was recalled as ha* the technical m o d e s of the kervit d u c t . All four technicians indicated kervit as a &ology that hada&cipated the main &mica1 develop ments thereafter established in the 1980s. And it is no accident that those technicians belonged, like Carnevali, to the generation that had been able diiectly to intaface with the kervit technology @ecause they had been employees of Ckramia Veggia or competing fkm.9). -t &ally, the tubular kilns used for firing majolica were also used in firing kervit Some eiaht metres in lenath. these had 32 to 48 chmels. The mouth of the kiln resembled so manv liGe windows, allif which were channels. The tiles moved forward because one pushed th;: other. The daily output of a tubular kiln was about 400-500 tiles. In the early 19509, kilns (known as "a pmso di pellegrino"), designed by engineer Drago in collaboration with Korach and Battistin had a production capacity similar to the tubular kilns. From the late 1950s on, however. Dal Borgo used m e 1 kilns that were built by the German firm Kerabedarf: they used small trolleys (approx. 66x66 cm), on which a single layer of tiles was placed, and their production capacity was around 300 square metres per day . In the 1960s a similar kiln, with a trolley for single layer firing, was proposed by Poppi for rapid single-fmng.
In the course of over twenty years, Dal Borgo developed numerous devices and tools for the k e~t production system8, but the only one which was widely used, and is still used to day, is what is h o r n as "the bell". As far as traditional twice-fire production is concerned, a process began in the early 1960s whereby moving operations were mechanked, and this process went on for over ten years. These were years in which a process of collective inventiont emerged. This process was made possible by a continuous flow of information between technicians and production experts, who moved from one f i r m to another, who were often personally acquainted and who shared in the common adventure of bringing to life Sassuolo's new ceramics induw.
Veggia, however, seemed excluded from this process: the kervit patent concerned a procedure which was not easy to hitate by introducing alterations so as to circumvent the patent protection, because infer alia employment of the kervit technology relied greatly on the tacit howledge acquired by D a l Borgo during the long stage of experimentation and production on industrial scaleS. Moreover, as is the case with many technologies, it was a question of developing complementary technologies, a * "At one point it emerged that with the pouring process, the tile did not come out flat, and this wasdue to the factthatthe tool we were u@g was theold orifice oftheglazingmachine which was made of a hopper which had an adjustable apemre at the bottom, which meant that a veil was formedfor the simple reason that pouring was slightly delayedbut in the middle it was slightly raised and overflowed towards the sides. Wen, I solved the problem in two ways.
Fmt using an instnunent, the bell, which they still use today. It was I who invented the bell to solve this very problem. Because, being tuned, less went into the middle and more went to the sides. But the problem of the edge remained, so I invented a little device: I applied a disc which acted like a knife and cut away 4 to 5 millimetres from a tile 15x15 cm or 20x20 cm, from each side in such a way that the tile was trimmed h g h t at the edges." [IDB 441.
But the pmduction of formats such as the mosaic also made it necessary to develop suitable tools. "The formats were basically 15x15 and 10,8x10,8 cm (4x4 inches, the American format); then when I set to thinking about the mosaic, I began producing 20x20 cm (it was clearly a little larger because the edges were later trimxned). The idea of the mosaic occurred to me like this: how do you cut glass? If you cut glass you use a diamond which makes a mark you then break. So I had this idea: I fitted very thin discs which had extremely sharp blades and were supported by springs strong enough to exert the required pressure to make a diamond incision in the unfired tile. These discs were positioned on the glazing line, and were stood on points so they could be dragged along: the tile passed by and the incision was made, that is, the line was t r a d in it". [IDB 491. When firing was over, the various pieces that made up the mosaic were broken off.
The belts themselves also needed to be modified. In particular "the belts on which the material was transporwl along the pouring and glazing lines easily got dixty and I had nails inserted in them to hold the tiles up higher" [lDB 1041.
t The notion of collective invention is discussed by Robert Men (1983). In a situation where the majority of firms did not allocate resources to research and development, collective invention was an efficient innovation process even for the individual f i r m .
$ "Patents covered the entire method and procedure throughout its development, nobody could have copied the technology, the invention was protected" [1DB122]. requirement which was to an extent amply fulfilled by Dal Berg's personal qualities, since he personally led the group of mechanics and carpenters'working under him at Veggia Nevertheless, the spread of automatic devices for moving operations in the traditional production process represented a challenge, both technical and economic, to which Veggia was unable to respond: the technical insularity which had developed inside the factory precluded the possibility of exchange with the outside world, and this acted against the interests of the kervit system.
In the early 1960s. the technical and economicnot to mention socialenvironment had come a long way fmm the days when the k e~t system was developed. There were now numerous firms producing majolica tiles which were a source of growing demand for specific mechanical innovations; not simply kervit adaptations. And this is one of the factorsthe other being the difficulty of imitating the processwhich explains the oblivion into which kervit fell following the closure of the Veggia plant. Oblivion here means that kervit technology was no longer adopted, butas a result of Dal Borgo's self exclusion fmm the process of collective inventionthe knowledge necessary to foster the many activities developing in the booming ceramic district did indeed emerge. In fact, Veggia's failure fixed human resources endowed with excellent technical training acquired in the long experience in kervit production. More open than Dal Borgo to the new social and technical opportunities emerging in the changed environment, many of the Veggia's ex-workers were in fact very active in starting up new inde pendent enterprises, as producers both of tiles and of ceramics machinery.

Innovation, complementary activities and industrial dynamics
The kervit experience left its mark precisely because it trained a vast quantity of technicians, but also because it was a model of technical innovation which was decidedly different from anything else in the district, as is witnessed by the radical technical alternatives it introduced in molding and firing, two vital areas of the production process. In particular, molding via pouring as used in the kervit system ran radically counter to the technical and economic interests of the press manufacturers* and the investigation * K e~t production certainly did not sweep away press production because, at all events, it continued to use refinctory plates, the mold onto which the preparation was poured, which were produced by a molding process using a press. Every plate was used in 80 or 100 production cycles and thus the potential development of press production would have been reduced to one hundredth.
of this aspect highlights some important liraks between innovation, the emergence of complementary activities and industrial dynamics.
&st of all, we should note that in the 1930s, when vitral was being developed, the two molding processespouring and pressingboth had the same degree of mechanization*. Molding by pouring increased the use of the glazing line: a piece of equipment consisting of a metal frame which supported the belt on which the tiles passed between one application and another. In molding by the use of presses, both the filling of the molds and the activation of the head which compressed the clay were done manually, and presses .had therefore a limited capacity, producing only small format tiles.
During the twenty years that saw the development of kervit, the refinement of the various machines involved in the pouring and glazing line resulted in higher product quality but did not substantially change the production capacity of that line, nor the degree of automation in the loading and unloading of the line, which was still done manually. Press production, on the contrary, underwent considerable transformation, leading to the ,development of powerful friction presses, which in the early 1960s were completely automatic. The increase in the level of mechanization had increased potential in terms of the capacity to press tiles larger in format than those available through the use of manual presses and, above all, it had increased production capacity, by thus requiring the development of semi-automatic and automatic devices for the extraction of pressed material and for Ioading the trolleys +.
The presses had always been produced by specialist h s , and by the end of the 1950s several Italian producers were making important headway in foreign markets, outstripping even the Germans, the original European market leaders. This was all thanks to production expansion within the Sassuolo district: Ceppelli (in Sassuolo), WeIko (whose head office was in Milan but which manufactured in Sassuolo) and Sacmi (of Irnola). These firms were much larger than other machinery producers with a product far more complex in terms of the components and the technical expertise required to manufacture them. The press producers initiated a lot of the * 'Ile notion of mechanization level refemd to in the text is that developed by Bright (1958).
t The increase in press production capacity showed up the inefficiency of manual performance of manual unloading of the materials because a larger number of workers would have been necessary for each press, and this was practically impossible owing to the physical space available at the end of the press. innovations related to the movement of material in various phases of the manufacturing process which they developed during the early 1960s as techniques complementary to pressing; and these innovations were produced either in house or by small engineering sub-suppliers from the district.
The search for solutions to various technical imbalances* between the many operations connected with the pressing phase was then a key opportunity seized upon by press producers to activate the all-important information exchange mechanismwith tile and components manufacturerswhich was conspicuous by its absence from the kervit environment.

ON TIIE CONCEPT OF GENERATIVE RELATIONSHIPS
Section 5 has highlighted some critical elements characterizing k e~t ' s demise, one might say: looking at actors initiating the changes in the agendamfact space in which kervit emerged. But it still remains to explain why no other company in the district decided to buy the k e~t licence and produce with k e~t technology after the closure of Veggia; and why the process of imitation that has been so fundamental a factor in the spread of many techniques in the ceramics district did not occur in this case.
A recurrent reply to this questionwhich is echoed in some of the interviews with ceramics expertstis that the main reason for abandoning kervit technology was the increase in labour costs during the 1960s. But the ethnographic analysis has allowed us to reconstruct the dynamics of the kervit innovation process, showing that the economic assessment remains generally valid but is insufficient to explain why k e .~t was abandoned.
My interpretation of the k e~t story highlights that other elements have played a major role in the innovation dynamicsi.e. the need to monitor and fuel the generative potential of relations in agents' and artifacts' space. It is within this framework that the kervit system remained unaffected by the process of collective invention, and the emergence of complementary activities was not exploited with the potentialities that other actors were * The role of technical disequiliium between the components of a machine and of a process is highlighted by Rosenbeg (1969) as being an important device in focusing innovative activity.
t Cf. Canani (1997). able to recognize. I should therefore l i e to conclude by focusing attention on three main issues around which to sum up this interpretation.

Monitoring and fostering generative relationships: the missing conditions
When Dal Borgo explains his decision to patent the procedures and devices he had invented, he highlights two basic reasons. The fust is personal pridethe pleasure of being recognized m m inventor* . The second is a strictly economic reason. Dal Borgo believed that the patent would guarantee financial gain from the invention. h d yet he did not deal directly with the matter since this required relational skills and intemational connections. Dal Borgo delegated this side to Korach, who was well connected in a variety of settings. Until the mid-1950s, in the relationship between Dal Borgo and Korach there were all the five conditions necessary to describe that relationship as generative of changes in the space in which they operated. However, though both acted in an aligned directedness (increasing kervit sales in the world market), differences in expertise and fields of activity that had positively marked the two actors' heterogeneity became a disadvantage because there was no longer mutual directedness. Korach, who took care of the commercial aspects of overseas licence sales, did not mobilize the technical and production linkages which might have been able to generate the development of kervit-related complementary techniquesnot just movement operations and &kg techniques, but also technologies and materials for fixing kenrid tiles.
At the beginning of his career as inventor in the 1930s and later as entrepreneur in the 1950s, Dal Borgo used his own technical ability to build around him a technology which was d i a d t to imitate and of which he intended to remain the sole user. From the supply side, there were few other companies which produced wall tiles using traditional production -* Although the technical and scientific community recognized his original contribution as the author of k e~t (cf. Vecchi. 1952;Korach, 1955). in the I d community Korach's fame was predominant. In this respect it must have been very galling to acknowledge over the years that the pleasure of being recognized as k e~t inventor was denied him because he wanted to share the patent with his friend and master hlaurizio Komch.
In the first intemew, D a l Borgo recalled that at the end of the war, K m h came into contact with the Reggio Emilia federation of the Communist Party. It was in this environment that the mmour spread that Korach was the true inventor of the kewit system, a m i m c e p tion abetted by the fact that Korach, apart from his fame ss an intellectual, was involved in preparing the patent registration documents and in selling the licence. Dal Borgo asked him to publish a denial in the local communist party magazine, but the article had little effect techniques and sold mainly to the Italian market. At the end of the 1950s, changes on the demand side brought about the growth of numerous firms which were soon to face the 1963-64 slump: dozens of firms closed, but those which managed to weather the storm gained momentum in the wake of the introduction of the tunnel kiln, high-output presses and the dramatic expansion in demand for floor tiles. In the 1960s. the economic and social climate was radically different from that in which Dal Borgo had started developing the kervit technology: alongside the hundreds of firms producing tiles there were now numerous manufacturers of specialist machinery and equipment. Though other ceramics firms also produced in-house some machinery used in the production process, the fact that, throughout the 1960s, Veggia continued to produce in-house all of the kervit specific equipment was certainly an exceptional situation. But this was no longer a choice vis-a-vis an environment which did not offer technical opportunities; rather, it was Dal Borgo who did not take account of changes which had taken place in the environment in which he was operating: he had always wanted to work alone and he thought he could continue doing so, without links with other technicians.
Some critical elements characterizing generative relationships were lacking. In particular, Dal Borgo no longer looked for interactions with heterogeneous agents, though in the past these had had such an important effect on his innovations. In the interviews he is quite clear on this point: according to him there were no other technicians in the firm with whom he could interface*. Nor did he seek contacts with outside: like every artisan, he felt himself to be in possession of a fundamental secret for "making a h e r , less expensive product". This view hindered him from taking up organizational hints even though they came from within the factory; and, basically, even the periodic meetings with kervit licensees acted as a means of transmitting to them the solutions he had adopted, but were never used to focus upon a joint initiative to bring about changes in kervit's position vis-&vis other, rival products. His overriding technical genius and his fixed view of ceramic production as artisanal production prevented him . from understanding that exchange of his with others' experiences might be fruitful. His centralization of technical decisions and authoritarian single-mindedness left others no time and space to talk, nor freedom of action in the common interest of Veggia (i.e. no right permissions). When agents interact, all these are essential conditions facilitating their understanding of their respective expertise and identity. And the possibilities which * Cf. 1DB69+70 and footnote, cf. also interviews with Lusuardi and Carnevati. emerge from their joint activities will be enhanced when their relationships are interwoven with a network of other relationships (Lane and Maxfield, 1997). The absence of right permissions had a negative effect on Dal Borgo's understanding of ongoing changes and helped to alienate him from the control of the technical and economic space he himself had created.

3 Technical and social dimension of the imitation process: collective invention and the emergence of new competences
What emerges from the kervit story is not only that k e~t made it possible to manufacture a wall tile that was difficult to produce, but also that it was hard to come up with anything similar, that is, inventing around the cover offered by patent protection. The difficulty in using this production method lay in the considerable variability of the technical parametersowing to the use of natural raw materials whose composition variedwhich were subject to a transformation p m s s in which environmental parameters (temperature and humidity) sig@ficantly alter the outcome of the process. The adjustments needed to achieve good results (in terms of a low reject percentage and uniform product quality) required a practical knowledge that could be acquired only with a great deal of production experience. Although this tacit knowledge could certainly help explain why kervit was abandoned, this explanation is not enough for other techniques had similar problems in subsequent years, but were not abandoned. In the case of kervit technology, we come nearer to an explanation if we also consider the technical and social isolation at Veggia: it is in this context that considerations of cost and an assessment of the massive necessity of tacit knowledge (to overcome technical difficulties) appear relevant.
K~M?S isolation was actually worsened by inter alia the impossibility of imitating it. From the kervit case history (but a similar case is that of Enduro technology patented by Marazzi in the mid-1980s) it would appear that imitation is a crucial factor for the success of an innovation: "many actors win, one alone loses". Even a far-reaching innovation cannot manage to assert itself if the monopoly position of its inventor proves a condition of weakness when the other producers are able to form an alliance against the monopolist, discrediting the reputed superiority of the proposed innovation. Moreover, potential users perceive the fact that there is only one supplier of the innovative product as a sign that "something must be wrong, otherwise the others would ha\ie done it"*. These two factors, which were to kervit's disadvantage, help to describe the technical and social isolation in which this technology found itself as compared with other alternatives that emerged in the 1960s.
The twenty years during which kervit production took place on an industrial scale have nevertheless left a legacy of knowledge which permeates the entire local production system. This is not only because Veggia trained large numbers of technicians, not only because kervit technology contained the essential ingredients for the technical development of the following forty years, but also because it actually provided an economic model to be avoided.
To appreciate this point, let us consider, for example, a different case which occurred in the 1970s, when Marazzi developed a single-fire production process which featured the use of a roller kiln. At first the technicians working on the new kiln were quarantined at the Fornovo pilot factory, far from the ceramics district and, when ready, the kiln was patented. Then, Filippo Marazzi threw open the doors of the Sassuolo factory (in which the new kiln was installed) to whoever wanted to see what was going on. It was not patenting the process, but its maximum accessibility which became the necessary condition for taking full economic advantage of the innovation. For it is through imitation that conditions may be created for the development of complementary techniques. This imitative process does not involve all agents globally -rather, it is a local process: you only imitate what is being done by your "neighbod. Proximity, here, is defined in spatial terms (you have to see it with your own eyes), but also in technical terms (you have to have similar skills to ' those of the person you are imitating), and relational terms. The latter means that the particular interaction in which observation of the anifact or of the process used by another agent takes place is made possible by the fact that between those agents there are other interactions on the personal and economic level. The general result of this imitation process is a repositioning of the overall system. This takes place because to imitate it is necessary to have similar skills, but agents are not identical and the imitator has specific competences and specific relationships with agents and artifacts. Such relationships can trigger new entities (new agents or artifacts), but also new competences: this is the outcome of the imitation process that * Similar considerations apply to many others filed (a well known case is the Macintosh story).
has characterized the innovation dynamics and the industrial dynamics in the ceramic tile district.

3 The social dimension of patents and dynamic complementarities
Dal Borgo, who only dealt with technical and operative matters, assessed the impact of endogenous change in demand on the economic environment in which he operated as secondary for his own activity. Other agents, however, saw those changes as the opportunity to modify the space relevant to their action, albeit within the limits of their technical ability. It was precisely these limits which guided the choice towards developing techniques easier to use than those in the kervit system but decidedly more remunerative in terms of the profits they were to make possibled.
We have seen in the case of kervit that patenting the production method can considerably influence the potential spread of the technology involved, but it might be objected that, basically, this story also clearly involves a question of scale. If k e~t production had been not a thousand but fifty thousand square metres a day, then it might well have necessitated the development of complementary techniques inside the firm or on the part of outside specialist producers. And this is one of the reasons why it does not seem opportune to consider patents as a reliable indicator of innovations achieved within a small business production sys-the individual f i r m is not big enough to develop in house all the complementary technologies which prove indispensable if its full potential is to be exploited, and neither does it always prove possible to find solutions to technical problems arising out of the use of a new technology. In such a context, by limiting the circulation of technical information, the patent does not therefore offer the chance of interaction between agents and &acts which could bring about the changes (new artifacts, new agents and new entities) required for the development of new technology. The importance of links with the outside world has been recognized as a crucial factor for the success of an innovation since Carter and Williams published their studiei in the late 1950st. However, what has emerged from our reiearch is that the key factor is not what is usually highlighted in  such studies, i.e. the link with the scientific world of the particular field in which the innovation is developed*; the key factor is the links which should have been created with the network of potential innovators of all the complementary technologies. It was the absence of these links which prevented kervit from becoming an alternative to the traditional majolica production system.