Wishing to emulate the advances made by their British colleagues (Stone Axe Studies), during the 1960s P.-R. Giot and J. Cogné worked to identify the main sources of the rocks used to create the Breton axes. They had success in identifying one particular dolerite, Type A metadolerite, which was sourced to a work area at Sélédin in the commune of Plussulien (Côtes-d’Armor) by C.-T. Le Roux in 1965. Another dolerite extraction site was subsequently confirmed at Le Pinacle in Jersey but, following this, work on sourcing the axes found in the west of Armorica was largely abandoned. This article revives the search and presents the results of the 2010 and 2012 fieldwork carried out on the Beulin dolerite axe factory at Saint-Germain-le-Guillaume (Mayenne). The north-western area of Mayenne is characterized by a swarm of dolerite dykes, which are orientated between NNW/ SSE and N/ S. Allowing for some uncertainty, the Beulin dyke dates from the Devonian/ Carboniferous boundary. Geographically its site is on a crest line rising to 129.5 m NGF (Niveau Général Français, ± equivalent to OD), the relief of which reflects the presence of the dyke. The rock on which the dyke intruded is Alexain Cadomian age Granite. The Beulin dyke in the area as a whole often takes the form of small domes on the surface but with the rock itself nonetheless buried quite deeply in the soil profile. On digging down the dolerite is eventually encountered as rounded core stones embedded in a dark brown clayey loam. The larger core stones are sometimes quite far apart with smaller ones in between, or separated by joints. At the field site the dolerite dyke is some 8.5 m wide and the dominant joints cutting the dyke have a different orientation, running NW/ SE. Within the dyke the least altered, freshest dolerite occurs as rounded masses with the rock becoming progressively more angular as more or less joint-bounded blocks occur. The discovery of the site resulted from following up on a surface scatter of flakes. Known 20th century agricultural working of the land had left gouges in the dolerite at a fairly constant depth of some 12– 13 cm. Gouges, probably from plough blades and suchlike toothed implements, can be as wide as 10-12 mm. Patterns of crossing marks quite possibly reflect the ploughing of these fields during the 1960s on N/ S and E/ W lines. The various traces had to be identified for what they are and separated out from those resulting from prehistoric workings. Over the cleared surface seven depressed areas correspond to recent stripping of the surface and some feature a resultant scatter of fresh fragments of rock. The form of the stripped areas is variable depending on the size and shape of the removed blocks. The loose associated sediment is aerated, rich in brown loam and contains a number of dolerite fragments. In addition, the outcrop situated in AB9-10 reveals very clearly the negative form of the blocks removed. Only part of the whole site, labelled 1, will be described here. One of the first layers in this section was US 9, which is located against the working face of the outcrop and is a very dark brown to black clayey loam. It contains numerous charcoal fragments, sometimes forming layers themselves, and also contains pieces of worked dolerite. This layer has evidence of use of the thermal shock method to extract the rock and is dated to 4815 ± 35 BP. Overlying US 9 is a dark brown loam containing centimetre-sized pieces with only the occasional larger element. This layer appears more complex since some squares had more broken fragments than others and, in addition, the dip of the lithic fragments indicated different accumulations. Elsewhere, at least one of these areas contained abundant charcoal with a grey-blue sandstone hammer. Carbonised fragments of probable apple or similar wood from this level gave a date of 4760 ± 40 BP which was supported by three others : 855 ± 30 BP, 4800 ± 35 BP and 4835 ± 35 BP. Layer US 5 was a light brown loam containing rare, centimetre-sized, joint bounded fragments plus rounded lumps, seemingly residual pieces of the bedrock. US 4 is a loamy orange sand, partly resting on US 5, and containing rare small reject fragments from the extraction process and the occasional worked piece. This level is interpreted as the spread of debris from an earlier extraction event working towards exposing the dyke on its west flank ; it is dated to 5660 ± 35 BP. US 2 covers the whole of the US layers. It is a brown loam with abundant fragments from the extraction process which is taken to represent the collapsed remains of this debris ; it is dated to 4890 ± 35 BP. US 1 contains a higher proportion of worked fragments and resulting spalls than the debris from the extraction process. The fragments are embedded in a light brown loam and may be considered a pile of reject spalls, an interpretation supported by the finding of a nucleus, placed in a flat position. Not all the dolerite outcrops were exploited. Numbers 1, 3, 5 and 6 are examples. It appears that the Neolithic people preferred the outcrops with a more widely spaced network of joints to those without. The favoured outcrops (numbers 2, 4, 7 and 8) were subjected to thermal shock treatment of their exposed faces— definitely in the closing stages of exploitation but almost certainly earlier on too. The aim was to enlarge the joints to allow the extraction of thicker blocks with a strike plane of 90° or more. One such block, fallen to the base of the working face, weighed about 150 kg and could be retrofitted to its original position. Carbonised fragments of probable apple or similar wood, discovered beneath this block, have a date of 5410 ± 30 BP, which places it neatly at the beginning of the Middle Neolithic II. Another block of about 200 kg was found and put aside for future analysis with the intention of obtaining pieces of dolerite that were thick and wide enough to allow the creation of short blades. From these first shapes, a bifacial form was fashioned by matching the edges, followed by trimming these to make the bevelled cutting edge and the narrow slightly pointed end. An accumulation of waste fragments, with a massive 66 kg nucleus at its centre, is the last stage of the sedimentation process and shows clearly that the fashioning of the axes occurred on site. Going beyond the discussion centring on the use of the thermal shock process (prior collection of wood specially chosen to create the strongest rise in temperature), the anthracological study has added to the data already known for the Mayenne département and considers the presence of beech. To complete the study, we open up the discussion to consider two issues. The first is the question of the diffusion of the axes, a line of enquiry that still has a long way to go, while the second focuses on the existence of the long rough-outs that probably served for the fashioning of the fusiform axes so well-known over the northern half of the Armorican massif.