As for cremations, the petrous bone was abraded to remove surface contaminants using a dental bur. The otic capsule was separated from the cremated petrous portions using a 1 mm mechanical saw mounted on a drilling machine (DREMEL® model 300), following Harvig et al.58 (link). Subsequently, the densest part of the central inner ear was sampled using a low-speed drill (2 mm diameter), producing clean samples of intact otic capsules for the Sr isotope analyses. The bone powder was stored in pre-cleaned plastic Eppendorf (1.5 ml) vials. The bone powder mass ranged between 0.02 and 0.04 g. As for inhumations, upper molar enamel was sampled from the protocone, or mesiolingual, cusp to the cement enamel junction (CEJ), whereas lower molars were sampled from the occlusal margin of the protoconid, or mesiobuccal, cusp to the CEJ, following Müller et al.54 (link). A flexible diamond-edged rotary wheel mounted on a drilling machine (DREMEL® model 300) was used to cut a longitudinal crown section of the cusps. Adhering contaminants such as soil, sediments and all trace of dentine were removed using a dental bur. The tooth enamel mass ranged between 0.02 and 0.04 g.
For 9 samples (from FM-SR-52 to FM-SR-60), a different methodology was adopted to sample bulk enamel, similar to the method proposed by Czermak et al.59 (link) for dental roots. This technique was employed to remove any dentine trace, which can be affected by diagenesis and can influence 87Sr/86Sr values contained in enamel samples (Supplementary Note). Teeth were cleaned and photographed. Subsequently, samples were covered in Crystalbond 590 Mounting Adhesive (Aremco Products, Inc.) before being embedded in resin. Crystalbond is a transparent resin, reversible in acetone, which isolates the tooth when it is embedded in epoxy resin.
Teeth were embedded in an epoxy resin and subsequently longitudinally cut following Nava et al. 60 . Sections were made passing through the tip of the dentine horn along the buccolingual plane. Once the cut was performed, dentine was thoroughly removed with a dental bur. The section was immersed in acetone to free the tooth from the resin (Supplementary Fig.S7 ).
For 9 samples (from FM-SR-52 to FM-SR-60), a different methodology was adopted to sample bulk enamel, similar to the method proposed by Czermak et al.59 (link) for dental roots. This technique was employed to remove any dentine trace, which can be affected by diagenesis and can influence 87Sr/86Sr values contained in enamel samples (Supplementary Note). Teeth were cleaned and photographed. Subsequently, samples were covered in Crystalbond 590 Mounting Adhesive (Aremco Products, Inc.) before being embedded in resin. Crystalbond is a transparent resin, reversible in acetone, which isolates the tooth when it is embedded in epoxy resin.
Teeth were embedded in an epoxy resin and subsequently longitudinally cut following Nava et al. 60 . Sections were made passing through the tip of the dentine horn along the buccolingual plane. Once the cut was performed, dentine was thoroughly removed with a dental bur. The section was immersed in acetone to free the tooth from the resin (Supplementary Fig.
Free full text: Click here
