Abstract:  Abstract : This study focuses on one of the contemporary innovations linked to the economy of academic publishing: the so-called transformative agreements, a relatively circumscribed object within the relations between library consortia and academic publishers, and temporally situated between 2015 and 2020. The stated objective of this type of agreement is to organise the transition from the traditional model of subscription to journals (often proposed by thematic groupings or collections) to that of open access by reallocating the budgets devoted to it. Our sociological analysis work constitutes a first systematic study of this object, based on a review of 197 agreements. The corpus thus constituted includes agreements characterised by the co-presence of a subscription component and an open access publication component, even minimal (publication “tokens” offered, reduction on APCs, etc.). As a result, agreements that only concern centralised funding for open access publishing were excluded from the analysis, whether with publishers that only offer journals with payment by the author (PLOS, Frontiers, MDPI, etc.) or publishers whose catalogue includes open access journals. The oldest agreement in our corpus was signed in 2010, the most recent ones in 2020 – agreements starting only in 2021, even announced during the study, were not retained. Several results emerge from our analysis. First of all, there is a great diversity of actors involved with 22 countries and 39 publishers, even if some consortia (Netherlands, Sweden, Austria, Germany) and publishers (CUP, Elsevier, RSC, Springer) signed many more than others. Secondly, the duration of the agreements, ranging from one to six years, reveals a very unequal distribution, with more than half of the agreements (103) signed for 3 years, and a small proportion for 4 years or more (22 agreements). Finally, despite repeated calls for transparency, less than half of the agreements (96) have an accessible text at the time of this study, with no recent trend towards greater availability. The analysis also shows widely varying degrees of openness, ranging from simple information on the ESAC directory through the provision of an open format to the allocation of a DOI and a reuse licence (CC-BY), including details of monetary amounts. Of the 96 agreements available, 47 of which were signed in 2020, 62 have been analysed in depth. To our knowledge, this is the first analysis on this scale, on a type of material that was not only unpublished, but which was previously subject to confidentiality clauses. Based on a careful reading, the study describes in detail their properties, from the materiality of the document to the financial formulas, including their morphology and all the rights and duties of the parties. We therefore analysed the content of the agreements as a collection, looking for commonalities and variations through an explicit coding of their characteristics. The study also points out some uncertainties, in particular their “transitional” character, which remains strongly debated. From a morphological point of view, the agreements show a great diversity in size (from 7 to 488 pages) and structure. Nevertheless, by definition, they both articulate two essential objects: on the one hand, the conditions for carrying out a reading of journal articles, in the form of a subscription, combining concerns of access and security; on the other hand, the modalities of open access publication, articulating the management of a new type of workflow with a whole series of possible options. These options include the scope of the journals considered (hybrid and/or open access), the licences available, the degree of obligation to publish, the eligible authors or the volume of publishable articles. One of the most important results of this in-depth analysis is the discovery of an almost complete decoupling, within the agreements themselves, between the subscription object and the publication object. Of course, subscription is systematically configured in a closed world, subject to payment, which triggers series of identification of legitimate circulations of both information content and users. In particular, it insists on prohibitions on the reuse or even copying of academic articles. On the other hand, open access publishing is attached to a world governed by free access to content, which leads to concerns about workflow management and accessibility modalities. Moreover, the different elements that make up these contractual objects are not interconnected: on one side, the readers are all members of the subscribing institutions, on the other, only the corresponding authors are concerned; the lists of journals accessible to the reader and those reserved for open access publication are usually distinct; the workflows have totally different objectives and material organisations, etc. The articulation between the two contractual objects is solely a matter of a financial distribution formula which, in addition to particular combinations between one an