1Between the end of the 19th century and the post-war period, the valleys of the French Alps underwent rapid industrialisation. The discovery of the hydroelectric potential of waterfalls, or 'Houille Blanche', celebrated by Aristide Bergès, led to the installation of various industries at the foot of the mountains between 1890 and 1930: electrometallurgy (steel, aluminium, ferroalloys), electrochemistry (calcium carbide, soda, chlorine, phosphorus, etc.), papermaking, and food processing industries (Blanchard, 1954; Veyret-Verner, 1949). Following a period of growth after the Second World War, since 1970 these Alpine industries have been in a state of decline, similar to that experienced by other formerly industrialised French territories (Combal, 2020). This process has led to a decline in industrial employment and the closure of factories (Bost and Messaoudi, 2017). This article analyses this period of deindustrialisation (1970–2025) through the concept of 'landscape trajectory'. Used in geography to study landscapes or territories over a relatively long period (Astrade et al., 2021; Forget, 2015), this concept is employed here to retrace, over fifty-five years, the swing between periods of intense transformation and phases of inertia that have characterised the productive landscapes of three French Alpine valleys. We hypothesise that, despite the general decline of heavy industry as an economic sector, the associated landscape trajectories are more varied and change over time. Comparative analysis of these trajectories allows us to question the concrete and symbolic place of industry in current mountain territories, regarding to what it has been in the past.
2The industry is present in many valleys in the Alpine chain. The article focuses on three French valleys in the Savoie department: the Val d’Arly between Ugine and Albertville, the Tarentaise between Moûtiers and Albertville, and the Maurienne between Villarodin-Bourget and Val d’Arc (Figure 1). These valleys are chosen for their concentration of past and present production sites in the electrochemistry and electrometallurgy sectors. They have also been the subject of an in-depth review in the 1970s by L. Chabert (1972; 1978), whose work forms the basis for our analysis of the period 1970-2025. The industrial sectors selected, sometimes referred to as 'large industry' (Chabert, 1978; Combal, 2020), are grouped here under the term 'heavy industry'. By this, I mean productions involving a high level of capital and equipment, located at the beginning of the production chain, and whose semi-finished products feed many other industries downstream.
Figure 1: Location of valleys and study sites
Background: IGN, 2025.
Created by C. Meynet, 2025.
3While the landscapes of hydropower (dams, penstocks) and the industries associated with it are historically linked in mountainous regions, this article focuses on the landscape trajectories of associated industries located at the bottom of valleys. Our attention is therefore focused on a particular aspect of the productive mountain (Perlik & Sega, 2022): the valleys and their industrial spaces. These spaces are usually located on the edge of waterways, in or near small urban centres, and at altitudes between 300 and 700 metres. These spaces, at the interface of 'urban and mountain models' (Sega, Viganò, Durand, in Perlik and Sega, 2022), are central in terms of function in that they connect the Alpine metropolises and the tourist mountain, yet are rarely present in the social imagination of the mountain (Bozonnet, 1992). While they are currently affected by the decline of their industrial function, these valley bottoms do not experience depopulation. Their economy is dominated by the so-called 'presential' or 'in-place' sphere (Davezies, 2008; INSEE, 2015). I use this terminology to refer to an economy dominated by services related to permanent and temporary presence in a territory. In the context of the Alpine valleys, I define the 'in-place' economy as one driven by the growth of the permanent population in the valleys and by tourism in the surrounding mountain ranges. Therefore, the studied landscape trajectories must be considered in a territorial context where industry is in decline but urbanisation is developing.
4In this article, the concept of landscape is taken in the sense of A. Berque (1984), as 'imprint and matrix', that is to say, as 'expression of a civilisation' (imprint) and as 'participation in the perception schemes' of a society (matrix). In the field of landscape studies, the concept of industrial landscapes is relatively recent, given that industry has often been considered a non-landscape (Edelblutte, 2024). It emerged at the beginning of the 21st century in a context where deindustrialisation could favour a process of recognition and patrimonialisation, not only of characteristic buildings such as factories, but also of peripheral productive components such as housing (Edelblutte, 2024). In this article, I understand productive landscapes as technoscapes (Okada, 2005), i.e. landscapes composed of technical installations dedicated to heavy industry production.
- 1 The analysis does not take into account the decision to close the French Ferroglobe establishments, (...)
5To achieve this, the methodology involved grouping and cross-referencing various materials, such as aerial photographs, company statistics and fieldwork. A census of aerial photographs taken by the Ordnance Survey (IGN) since 1970 was carried out using the 'back in time' tool ('Remonter le Temps') to identify changes in the landscape on eighteen sites studied: ten sites that closed successively between 1970 and 2021, and eight sites that were still in operation in 20251. For some of these sites, the aerial photographs are supplemented by photographs extracted from the work of L. Chabert. The interest lies in comparing them with current photographs taken from the same viewpoint, in a diachronic perspective. These landscape trajectories are put into perspective with the help of data from companies operating or having operated on each site. The socio-economic impact of industrial decline is considered using two non-exhaustive indicators: the evolution of employment in the selected industrial establishments, and the number of establishments that have closed since 1970. The number of employees at the beginning of the period is taken from the work of L. Chabert (1972; 1978). For active establishments and those closed in the last decade, data from the State's Business Directory is used as a basis and refined by monitoring the local and specialised press (Le Dauphiné Libéré; L'Usine Nouvelle; Les Échos). The chronology of closed establishments is established on the same basis.
6Several terms are discussed in light of the terrain in order to characterise the landscape trajectories reconstructed in this article, whether they concern active or closed sites: inertia, transformation, erasure, dilution and reindustrialisation. Using the detailed economic indicators mentioned above, the article questions the adequacy or inadequacy between the socio-economic trajectory and the landscape trajectory of the industrial sites studied. This discrepancy is discussed in terms of the inertia/transformation dichotomy: I draw on the work of D. Buu-Sao and L. Patinaux (2023) on the inertia of transition. However, this is not an opportunity to discuss the inertia of an economic system such as capitalism in its industrial form. Our focus is on the inertia of productive landscapes, and on how this inertia – or its absence – reveals the actors and issues present on each site. Another landscape trajectory that I analyse is erasure: the majority of production sites closed between 1970 and 2025 have been erased from the landscape, in whole or in part. The article then seeks to reconstruct the timeframe and modalities of this landscape erasure. The concept of dilution (Edelblutte, 2024) is used to describe certain productive landscapes that are still present, whose forms become blurred in the growing urbanisation of the valley bottoms. I hypothesise that this is another form of erasure. Finally, other trajectories emerge, particularly in the case of recent closures: two cases of the reindustrialisation of former sites are analysed par their landscapes. I then question the form that this political objective, as set out by the French state (Voy-Gillis, 2022), can take locally.
7Figure 2 shows ten establishments that closed between 1970 and 2025: seven in the mettalurgical sector and three in the chemical industry. In terms of manufacturing, aluminium is the most represented in closures (four establishments), ahead of steel and calcium carbide (two establishments each). Geographically, these closures mainly affected the Maurienne (six) and the Tarentaise (three). This represents 60% of the establishments in operation at the beginning of the 1970s in the Maurienne valley, and 50% in the Tarentaise valley. This trend is not unique to the Alpine valleys and is part of the deindustrialisation affecting all French industrial areas. Among the factories that closed during the 1970s and 1980s, it is often the case that they were located far from the main urban centres in the local and regional areas. In Maurienne, the sites at La Saussaz, Prémont, La Praz and Villarodin-Bourget are the most upstream. Another factor that could explain some closures is the topographical constraints of the sites and the impossibility of expanding: this was the case at La Saussaz, where the factory was constrained by the river Arc, the railway and the hillside. However, these geographical factors are not sufficient: the nature of the products manufactured, sector-specific technical innovations and economic trends have also contributed to some closures (Chabert, 1978; Combal, 2020).
Figure 2: The decline of industry (1970–2025) in three Alpine valleys: Val d'Arly, Tarentaise and Maurienne
Sources: Chabert, 1978; State’s Business Directory, 2025; Press.
Background: IGN, 2025.
Created by C. Meynet, 2025.
8Table 1 summarises the decline in staff numbers between 1970 and 2025 for establishments that remained active. On average, this decline is 45% for all establishments in 2025, but this average hides significant disparities, with some establishments experiencing moderate declines in personnel between 1970 and 2025, such as the Montricher and La Chambre plants in Maurienne, while others have seen a 75-80% reduction in personnel, with occasional mergers of establishments during this period (La Léchère), or conversely, the division of an establishment into several entities (Ugine). Based on this observation, I will now examine the landscape trajectories of sites that remained active between 1970 and 2025, followed by those that closed.
Table 1: Evolution of employee numbers in active establishments between 1970 and 2025
Sources: Chabert, 1972; State’s Business Directory, 2025; Press
Created by: C. Meynet, 2025.
9The landscape trajectories of active industrial sites are analysed based on what I refer to as the site envelope, i.e. the built elements and storage spaces included within the closed perimeter of the establishment. Table 2 shows the changes in the sites extents and compares them with the evolution of employment over the period. For the most part, the extent remain stable or shrink slightly, while employment decreases in all sectors. The most notable case of decoupling is that of the Tokai Cobex carbon and graphite plant in La Léchère, where the extent remained stable while the workforce decreased by 75% between 1970 and 2025. A similar trend can be observed at the aluminium plant in Saint-Jean-de-Maurienne, which doubled in size between 1972 and 1986 in two successive phases (see figure 3), while employment decreased by 25%. However, this decoupling is only apparent: the expansion must be considered in the context of the closure of the two other aluminium plants in the area during the same period: La Saussaz and La Praz. The decline in employment occurred after this expansion. In other sites, the evolution of employment and the built area are in the same direction, i.e. a decline in employment corresponds to a spatial schrinkage of the site. In Épierre, the closure of the phosphorous furnace in 1995, a long hall practically 200 m long, marked the end of a decline in employment that had been underway for about twenty years. The demolition of this building in the following years, the most striking feature of the landscape, reduced the phosphorous plant in Épierre to a small site with a small workforce (Table 1).
Table 2: Evolution of the built-up area of industrial sites that remained in operation between 1970 and 2015
Source: IGN aerial photographs – Remonter le Temps
Created by: C. Meynet, 2025.
Figure 3: The aluminium plant in Saint-Jean-de-Maurienne: gradual expansion and integration into the urban landscape
Photographs: L. Chabert (1972 and 1987); C. Meynet (2025).
10Another evolution has occurred through the densification and centralisation of certain sites, at the expense of workers' housing. In two establishments, Ugitech in Ugine – formerly Pechiney – and Arkema in La Chambre – formerly Elf-Atochem – workers' housing located either within the production area or in the immediate vicinity is demolished and replaced by industrial buildings or production annexes. In La Chambre, two residential buildings located within the site's perimeter were converted into a building dedicated to the site's emergency and security services (site visit, March 2025). In Ugine (Figure 4), the phalanstery, an emblematic building of the social utopia pursued by the founder of the Girod steelworks, Paul Girod, is partially demolished. Built in 1910 by architect Maurice Braillard, its U-shaped design incorporated family housing in the two wings and single-person housing in the centre, offering a high level of comfort for the time (Bal, 1993). The two wings were demolished at the end of the 1990s to make way for an extension to a neighbouring industrial building. Furthermore, several bachelor buildings and a workers' estate with basic amenities had been demolished and replaced by industrial buildings before 1970. Taken together, these changes reveal the decline of working-class housing – at least in its oldest and most basic forms – in or near production sites. They also suggest changes in the composition and aspirations of the working class: more comfort and a less immediate location for residences in relation to the factory.
Figure 4: Site production densification at the expense of workers' housing: the example of Ugine
Photographs: IGN – Remonter le Temps.
Modifications: C. Meynet, 2025.
11A final observation regarding certain active industrial sites is their dilution within the urbanisation of the valley bottom (Edelblutte, 2024). Alpine valleys are now characterised by the coexistence of three main functions – living, production and circulation – in a hybrid industrial and urban landscape (Sega, in Perlik and Sega, 2022). The aluminium plant in Saint-Jean-de-Maurienne (Figure 3) illustrates this idea of hybridisation and dilution: whereas until the 1980s the plant was clearly separated from the town and the surrounding area was sparsely populated, dilution has occurred since then through the development of activity zones on the outskirts of the site, and more recently through the construction of the Lyon-Turin base tunnel.
12Motorways and fast road can also contribute to the dilution of the industrial landscape through their layout. Indeed, their route often takes them away from certain factories that were previously located on the edge of the road. In Tarentaise, for example, the construction of the dual carriageway at the beginning of the 1990s put certain industrial sites at a distance, such as the Bâthie factory, which is located on a lateral alluvial cone and not in the valley bottom where the fast road passes. The motorway in Maurienne also veers away from the Arkema factory in La Chambre (Figure 5), unlike the main road and railway line which run alongside it. However, other factories are crossed or skirted by these major transport routes, such as the two factories in La Léchère, divided by the fast road, or the aluminium factory in Saint-Jean-de-Maurienne (Figure 3), skirted by the motorway for 1.5 km. In these cases, the industrial landscape is at the forefront of the journey. However, by accelerating mobility, these motorways and fast roads blur the perception of the landscape in general, and of industrial landscapes in particular: speed becomes another factor of dilution of the landscapes crossed (Nageleisen, 2007).
Figure 5: The La Chambre factory (Maurienne) in the industrial and urban landscape of the valley bottom
Photo and interpretation: C. Meynet, 2025.
13Between 1970 and 2025, ten heavy industry sites closed on the three sites under study. Figure 6 compares their current state with their previous state, without considering the temporal aspect of the transformations undergone, which I will address later. Five situations can be identified, three of which involve the erasure of industrial traces between the closure date, i.e. the date when production was stopped permanently, and today. Of the ten closed sites, seven have been demolished, and only three have been preserved or partially preserved. Of the seven sites that have been demolished, six have been covered by another land use. Only one demolished site has not been covered, the chlorine plant in Prémont in Maurienne. The sites that have been covered present a mixed coverage: the footprint of the former factory corresponds to built-up areas, parking and storage of materials today. Two factories have been covered by roads: La Praz and La Saussaz are mostly located under the A43 motorway. The two sites that were shut down in the last decade are the only ones to have retained their general built-up extent.
Figure 6: Current state of the ten sites closed between 1970 and 2025
Sources: IGN aerial photographs and field observations
Background: IGN, 2025
Realisation: C. Meynet, 2025.
14If I now look at the post-closure period in more detail, a wide variety of landscape trajectories emerge, which Figure 7 compares. Four types of trajectory emerge.
Figure 7: Landscape trajectories of the ten sites closed between 1970 and 2025
Sources: IGN aerial photographs, fieldwork, press.
Created by: C. Meynet, 2025.
15Trajectory A stands out due to the contrast between an inert phase stretching between ten and fifteen years, during which the closed sites are left in place, and a rapid demolition and transformation phase (approximately five years), resulting in a stabilised new use. The three factories concerned are the ones located furthest upstream in the Maurienne valley, and among the oldest closures. La Saussaz and La Praz have practically identical trajectories, which can be explained by their comparable location on the edge of the river Arc and on the A43 motorway. It was the construction of the latter at the end of the 1990s that brought an end to the inertia phase and accelerated the transformation of the two sites. The only difference is that La Praz underwent demolition in two stages, while La Saussaz was demolished in 1995, just before the construction of the motorway (Figure 8). The Villarodin-Bourget plant was not affected by the motorway as it was located upstream of the Fréjus tunnel. Demolition began in the early 1990s, almost twenty years after its closure, and gave way to a small business workshop area which occupied only part of the former site for twenty years. During the 2010s, the site underwent a second transformation when it became a base for the Tunnel Euralpin Lyon-Turin construction.
Figure 8: The landscape trajectory of the Saussaz site (Maurienne): from the factory to the motorway.
Photographs: IGN – Remonter le Temps.
Arrangement: C. Meynet, 2025.
16Trajectory B encompasses three sites where the post-closure inertia is shorter than in Trajectory A, but the landscape transformations were more abrupt. The three affected factories have in common the fact that they closed in the 1990s and remained relatively outside the scope of transformation projects. The Prémont factory was demolished four years after its closure, but it has not found a new use. The cleared site has been left exposed and vegetation has grown back spontaneously since then. This is probably due to its location slightly on the slope and not in the redeveloped valley for the motorway, unlike its neighbours in La Praz and La Saussaz. The Venthon factory (Figure 9), closed in 1994, was mostly demolished soon afterwards, except for a few buildings that were kept until 2002 for research and development activities of the neighbouring Ugine factory – now Ugitech. The second phase of demolition took place between 2003 and 2006, but the site remained unused until the early 2010s, resulting in spontaneous growth of vegetation. It was finally the Arlysère intercommunal community that, during the 2010s, turned it into an 'ecoparc', i.e. a waste disposal site dedicated to professionals, local authorities and associations, which mainly accepts large and bulky waste. In La Pouille, the main manufacturing building was demolished before the permanent closure of the establishment. However, the period after closure is marked by a phase of inertia lasting around twelve years, during which time no demolition work is carried out on the three remaining buildings. One building was finally demolished and replaced by a site for small business workshop between 2009 and 2011, but the other two buildings remained derelict.
Figure 9: The landscape trajectory of the Venthon site (Val d'Arly): from the factory to the waste disposal site
Photographs: IGN – Remonter le Temps.
Arrangement: C. Meynet, 2025.
17Trajectory C stands out due to the speed of post-industrial transformation in a context of significant land issues. The Moûtiers and Villard-du-Planay factories have in common their attractive location: in a small town and adjacent to the station for Moûtiers, and in a much smaller hamlet but close to tourist resorts for Villard-du-Planay. In both cases, demolition and functional transformation follow quickly after closure. The Moûtiers factory (Figure 10), located in a restricted and terraced site, was demolished in two phases: the upper part was demolished as soon as it closed in 1982, and the lower part was demolished between 1988 and 1990 under pressure to accommodate the Albertville Winter Olympics. Indeed, the site was chosen to host the International Radio and Television Centre (CIRTV) during the event, which was built between 1990 and 1992. After 1992, it was gradually converted into a mixed zone combining shops, small business workshops and housing. The Villard-du-Planay factory was demolished in 1990, three years after its closure, and transformed in the following decade into a mixed zone of small residential buildings and small business workshops, which has become denser since then.
Figure 10: The landscape trajectory of the Moûtiers site (Tarentaise): from the steelworks to the current mixed use of activities and housing
Photographs: L. Chabert, 1970; C. Meynet, 2025.
18Trajectory D corresponds to the two factories that closed in the last decade. Although I lack a certain amount of perspective in our analysis, I can still make a few observations. In Maurienne, the steel foundry in Saint-Michel-de-Maurienne, on the Temple site, closed permanently in 2016, after a long period of decline that began with the closure of the steelworks in 1977. In Tarentaise, the Ferroglobe ferroalloy plant in La Léchère, on the Château-Feuillet site, closed at the end of 2021. In different contexts, these two production sites are currently undergoing various stages of reindustrialisation in situ, over a relatively short period of time. In Saint-Michel-de-Maurienne, a period of inertia elapsed between 2016 and 2019, before a portion of the premises (10,000 square metres) was purchased by a local metalworking company seeking larger premises. However, it remains a small employer with 30 staff. In Château-Feuillet, the reindustrialisation project is almost immediate: Ugitech, a company based 50 km north, expresses interest in buying the site in 2022 as part of a project called Ugi'Ring. Ugitech plans to adapt the existing facilities to create a recycling plant for industrial waste containing metal elements (nickel, manganese and zinc). These elements must be melted down into alloys and reintroduced into Ugitech's steel production process, thereby reducing its reliance on external sources of ferroalloys. The two examples differ in terms of their landscape trajectory: in the first case, the reindustrialisation of the site was carried out without any landscape disruption: no changes are visible other than the new entrepreneur's plaque on the façade. In the second case, the reindustrialisation requires partial transformation of the site, which has been underway since the end of 2024 (Figure 11). The areas that have been demolished or are in the process of being demolished currently form a constantly changing landscape.
Figure 11: The Château-Feuillet factory (Tarentaise), undergoing transformation
In red: the sectors that have been or are in the process of being demolished.
Photograph: C. Meynet, 2025.
19Studying the trajectories of productive landscapes in a context of deindustrialisation leads to several conclusions.
20Firstly, the deindustrialisation occurring in the studied valleys is generally characterised by the erasure of productive landscapes. This erasure takes several forms and is not always synonymous with the disappearance of the industry. It can be the result of factories that close and are demolished, on various time scales and not always completely. Erasure can also occur through dilution: in the urban sprawl that occurred over the last fifty-five years, the remaining productive landscapes have often become part of an industrial-urban landscape, blurring them and making them less identifiable. Finally, in some cases, erasure occurs through the separation of industrial sites and transport routes, and the acceleration of mobility, which reduces the time during which productive landscapes are visible when crossing the valley bottoms.
21A second observation is that productive landscapes that remained active between 1970 and 2025 are relatively stable, without being fixed: some sites have expanded, others have shrunk, and others have become denser within themselves. It can be hypothesised that some establishments are currently out of proportion to their size, when the number of employees has fallen sharply for a site that has remained the same size. Another hypothesis is that industrial sites are places that regularly reorganise themselves in response to innovations in production processes, market fluctuations and new opportunities, or storage space requirements. All of this takes place within the installations and is not visible in the landscape: the buildings, which are often old, can create the erroneous impression of archaism, even though the technologies used to produce steel, silicon or solvents within them are sophisticated and evolving.
22A third observation is about the landscape trajectories of closed sites over time. While these trajectories are diverse, they are all characterised by the need for space and the existence or absence of a project driven by industrial and territorial actors. When a project exists, it accelerates the transformation and determines the new use. Without a clear project, the site either remains in place in a state of more or less deteriorated built-up wasteland, or it is demolished without being reallocated for another use. As M. Modica (2022) points out with regard to the reconversion of industrial brownfields, its effective realisation depends more on the territorial context than on the content of the brownfield. Another lesson can be learned about the dynamic and reversible nature of past productive landscapes: a new use at a given time does not preclude subsequent changes in a relatively short period of time, as was the case in Moûtiers.
23Finally, and to reprise A. Berque's (1984) dichotomy, the productive landscapes studied in this article are a fine but partial, and often time-delayed, imprint of deindustrialisation. Conversely, the erasure of landscapes and the lack of heritage conservation of buildings that dominate the trajectories of closed sites contribute to making the landscape the potential matrix of a certain social forgetting – passive or intentional – of the industrial past and present of mountain territories.