Known for more than 150 years, Cabeço da Amoreira is part of the Muge Mesolithic shellmiddens discovered by Carlos Ribeiro in 1864 along with Moita do Sebastião and Fonte do Padre Pedro (Ribeiro, 1884), only 1 year after locating the first two middens, Arneiro-do-Roquete and Cabeço da Arruda (Cardoso and Rolão, 1999/2000). Apart from two of the middens, Cabeço da Arruda and Cabeço da Amoreira, which have been excavated, first by Ribeiro's team in the 19th century, then by Mendes Corrêa in the 1930s, by Roche and Veiga Ferreira in the 1950s and 1960s, by Rolão from 1999 to 2001 and, since 2008, by Nuno Bicho's team, the other sites were completely destroyed by agricultural activities. Located in the Tagus valley, more precisely in the Muge region (Fig. 1), the site is one of the largest mounds in the area, exhibiting an elliptic shape mound of c. 60 × 40 meters in diameter and approximately 3.5 meters height at its centre, mostly composed of very large concentration of shells (Bicho et al., 2011). Like all other middens in the region, it stands on a lower plain and is approximately 15 meters high. Its southern exposure takes advantage of sunlight during most of the day (Gonçalves, 2014). According to the large extension of the sites around the Muge area, and the incredible number of artefacts, shells and its more than 300 human burials, it has been stated that these shell middens represent extensive residential occupations associated with long term stays (Arnaud, 1986, Arnaud, 1987 and Arnaud, 1993) though Rolão (1999) refers to the fact that there might be two types of sites: the large ones (Moita do Sebastião, Cabeço da Amoreira e Cabeço da Arruda) that could represent residential loci; and smaller middens (Vale da Fonte da Moça and Cabeço dos Morros) that may be associated with seasonal and/or logistical functions. Recent studies (Bicho et al., 2010 and Bicho et al., 2011) revealed that the Mesolithic people from Muge had important stylistic differences, suggesting the presence of regional social differentiation: while their material culture and economy are identical within the region, there are differences in lithic materials characterized by a higher presence and frequency of geometric microliths at some sites (Cascalheira et al., 2015 and Jesus et al., in press) as well as in the origin of both animal and plant diets (Bicho et al., 2013, Umbelino, 2006 and Umbelino et al., 2007).

According to radiocarbon dating performed on human remains, shells and charcoal samples from different areas (Bicho et al., 2011 and Bicho et al., 2013), the shell midden (Fig. 2) was used from c. 8000 cal BP up to 6600 cal BP (with few and short interruptions), and its abandonment coincided with the increase in soil aridity and the end of salinity of the rivers that compose the Tagus valley (Bicho et al., 2010 and Bicho et al., 2013).

Several habitat features found under the shell deposits, such as hearths, post-holes and different types of storage pits may indicate that the site was first used as a residential camp, from c. 8000 to 7900 cal BP, before the accumulation of a shell midden that took place between c. 7800 and 7600 cal BP. The presence of ceramic fragments on top of the midden, some of them bearing decorative features, reveals a Neolithic occupation which is supported by radiocarbon dates from two skeletons found on the top layers (7146–6847 and 7570–7320 cal BP), along with isotopic analysis that revealed an absence of marine dietary indications on one of the individuals (Bicho et al., 2013).

The identification of both Mesolithic and Neolithic contexts was the main reason for opening several test pits around the main shell midden area, including in the 2010 field season a 1 × 12-meter long trench (Fig. 3), which allowed the identification of five archaeological layers, formed during and after the shell midden, containing thousands of lithic artefacts, several ceramic fragments and two hearth pits (Cascalheira et al., 2015). The collection of shell ornaments used in this study comes from this trench.

The first time that ornamental artefacts (bearing artificial perforations) from Cabeço da Amoreira were mentioned was at the XV Congrés International d’Anthropologie et d’Archéologie préhistorique, held in Paris in 1931. Mendes Corrêa (1933) acknowledged the presence of several perforated gastropod shells from both salt and freshwater environments such as Cypraea europaea, Nassa reticulata, Bythinia tentaculata and Neritina fluviatilis, presently known as Theoduxus fluviatilis (Montfort, 1810), as well as other personal ornamental items including those made from animal teeth (e.g. Cervus elaphus) and small clay plaques. Later, Jean Roche (1966) addressed the ornamental ensemble found in the shell midden by reporting for the first time the presence of perforated items made of Pecten maximus and Cardium norvegicum alongside quartzite, schist and mammal bone pendants, some of them bearing visible wear marks. The variety and quantity of gastropod species, together with the amount of different raw materials that compose the ornamental set from the midden, led Rolão (1999: 197) to state that it is “the most complete set of all the (Muge) shellmiddens known to this date”. Apart from those previously referred, perforated specimens of Trivia europaea, Ostrea sp., Scrobicularia plana and Cardium edule comprise the more than 2000 shells altered by humans, recovered from the 19th and 20th centuries past excavations with little or no provenience information and thus, no data on chronology, stratigraphy or associated features and burials. According to Rolão's description, Theodoxus fluviatilis is the most common gastropod species at Cabeço da Amoreira with c. 1600 specimens, followed by Trivia europaea and Nassarius reticulatus, with c. 300 and c. 100 specimens, respectively. No technological or taphonomic information has been published for these specimens.

Since 2008, the projects entitled “The last hunter-gatherers in the Tagus valley: The Muge Shell middens” and “The last hunter-gatherers of Muge (Portugal): the origins of social complexity”, both funded by the Portuguese National Science Foundation (Fundação para a Ciência e Tecnologia) to Bicho, have recovered perforated gastropod shells, along with other materials such as mammal teeth and bones, and clay beads have been found during excavations of the Cabeço da Amoreira shellmound. These have not yet been analysed, but we expect to perform this study starting in 2016. The macroscopically observation of the set tends to reveal that these items were altered and are likely similar to those referred to in previous publications.

A total of 21 items bearing ornamental characteristics, such as perforations, were recovered from the Trench area during the 2010 excavation campaign (Bicho et al., 2011 and Cascalheira et al., 2015). The set is composed of 19 non-edible gastropod specimens, 15 specimens of Theodoxus fluviatilis and 4 Trivia sp., 1 cervid tooth and a fragment of a clay ring. The majority of the items were found in the Mesolithic shell midden context (layers 5, 3 and 2), except the clay ring fragment that comes from layer 7, identified as Early Neolithic, and 1 specimen of Trivia sp. that comes from layer 8 (Table 1), a non-archaeological layer (Cascalheira et al., 2015). The sample is mostly well preserved, which made its identification easier, apart from the 4 specimens of Trivia which had their top surface absent, allowing only the identification of the genus.

As stated by Álvarez Fernández (2010), the presence of small molluscs with no dietary importance at archaeological sites might indicate that these were collected with the main purpose of being altered and used as ornaments. The holes as well as the surfaces were microscopically analysed with a Lumar V12 Stereo-Microscope from Zeiss, equipped with an AxioCam MRC, which allowed us to determine the origins of the perforations and its wear characteristics, perform taphonomic analysis and gather all the perforation measurements.

Theodoxus fluviatilis (Monfort, 1810) is a small, oval freshwater gastropod that can be found in rivers and springs, preferably with high levels of calcium like those found in limestone bedrock environments. When present in shallow waters, they may reach a maximum diameter of 10 mm (Nobre, 1941), although they can also adapt to high salinity waters near the Atlantic coast (Nicol, 1938). Its shell has a high chromatic and pattern variation, which might be one of the reasons, together with its availability around the Muge area, for the high quantity of specimens found in Cabeço da Amoreira.

Five different species comprise the genus Trivia, present in both Atlantic and Mediterranean shores of Iberia (Alba et al., 2001). Commonly known as cowrie (Graham, 1988:326), its natural habitat is from the intertidal zone down to circalittoral (Toreli, 1982), but they can also be found post-mortem on the sandy shores near the rocky areas where they lived.

Two types of species, T. arctica (Pulteney, 1789) and T. monacha (da Costa, 1778), are very alike and, despite the differences in their coloration, they were once thought to belong to the same species, Trivia europaea (Montagu, 1808). The only difference between the two shell species is the presence of three dark dots on the top surface of T. monacha, visible only on adult specimens. Both species have solid elongated shaped shells with a narrow and slightly curved aperture, and the outside surface is characterized by the presence of striations (Pelseneer, 1926). Concerning the dimensions of the Iberian population, the adult specimens of T. arctica can grow to a maximum of 12.75 mm (Alba et al., 2001) and the T. monacha specimens can achieve 15.4 mm (Pelseneer, 1932).

Commonly known as red deer, Cervus elaphus (Linnaeus, 1758) was the large mammalian species most hunted by the Muge Mesolithic population (Detry, 2007). Though it is a common species in Europe and Asia, living in various temperate climate habitats such as flat lands, prairies or mountain landscapes, in Portugal, due to demographic pressure, they are today confined to specific regions. The clay ring fragment recovered from the study area has a semi-circular shape, with round edges, and is characterized by the presence of a hole in its centre.

According to several studies of experimental manufacture of shell beads in both Paleolithic and Mesolithic contexts (Álvarez Fernández, 2006, Benghiat et al., 2009, Cabral and Monteiro-Rodrigues, 2015, Francis, 1982, Stiner et al., 2013, Taborin, 1993, Tátá et al., 2014 and Vanhaeren et al., 2013), there are six different methods that can be applied in shell perforation and can be used to identify the techniques with morphological results similar to those from archaeological contexts. These are direct pressure, direct and indirect percussion, rotation, scratching and abrasion, performed with the aid of several lithic, bone and antler pointed implements manufactured specifically that, when combined with different movements, can reproduce a wide range of perforating shapes (Table 2).

According to the experimental procedures performed in the Theodoxus fluviatilis specimens, the better results were obtained by three techniques: •

rotation with lithic implements on the exterior surface – this technique results in a highly patterned, very circular hole with a regular contour and frequent external wedging;

Regarding the Trivia specimens, its morphology prevents internal perforation due to the very small and narrow opening and also, in result of the shell shape and size, indirect percussion was not attempted. Nevertheless, the experimental procedures using other techniques such as rotation, direct percussion and scratching revealed that perforation is very easy, as long as a lithic implement is applied. While rotation is the technique that produces the most perfect holes, direct percussion is the one that results in a less wear for the lithic point.

The formal analysis of the perforations was carried out only on those shells that were complete, which comprises 10 of the 15 specimens of Theodoxus fluviatilis and the cervid dental piece.

Most of the Theodoxus fluviatilis specimens recovered from the Trench area are intact and exhibit single holes (Fig. 4a) apart from 5 that were missing the top section (Fig. 4b), either consistent with the perforation action (Álvarez Fernández, 2006 and Tátá et al., 2014) or caused by post-depositional actions. The external surface of the shells show some degree of discoloration, possibly due to the high levels of calcium carbonate present in the midden, but the original patterns are still visible in most cases and, apart from the 5 broken specimens, all maintain their original shape which means they were collected alive or immediately prior to death, on the river shore (Dupont, 2006 and Dupont et al., 2014).

Concerning the holes themselves, these are regular with both circular and oblong shapes (Fig. 5a), and irregular shaped ones (Fig. 5b), with a 50% ratio. The circular or semi-circular shaped holes are characterized by a regular contour with soft exterior edging, while the irregular shaped ones are more jagged and external wedging is present in all five specimens. Regarding the size of the perforation, the average diameters are slightly higher on the regular and circular shaped holes (Table 3). These characteristics are not consistent with those present on holes that were naturally perforated by predators, with its almost perfectly round and smaller (< 2 mm) holes (Cabral and Monteiro-Rodrigues, 2015).

The perforation morphology varies, as mentioned earlier, but according to the results of the experimental procedures on shells from the same specie (Tátá et al., 2014), both the rotation with lithic implements on the exterior surface as well as internal direct pressure with bone or antler perforators with a cork anvil were likely used. The regular contours of the holes, the presence of external wedging and no internal flaking visible on half of the specimens are consistent with the rotation technique applied on the external surface using a lithic implement. This method requires that a controlled pressure is applied on the shell at the same time the rotation movement is made. The techniques used to perform the holes on the rest of the set, which have less regular or even irregular contours and exhibit exterior wedging and no internal flaking, are consistent with internal direct pressure using bone or antler perforators. Both technique and implements used to perform these holes make it difficult to control their size and morphology.

The perforated dental specimen is an inferior incisor (Hilson, 1992 and Schmid, 1972) and is characterized by two circular bi-conical shaped holes, with accentuated wedging on both sides of the root (Fig. 6).

The size of the perforation diameters does not differ much from one side to the other, ranging from 3.3 up to 3.5 mm in the maximum axis and 1.6-1.8 mm in the minimum axis. This bi-conical shaped hole was performed by rotating a lithic implement and, at the same time, by applying pressure. This technique had to be performed on both sides of the tooth in order to get this particular morphology.

Even though the entire set of Trivia specimens, composed of 4 artefacts (Fig. 7), are missing their top surface, it is possible to observe traces of what might have been two holes, on both extremities of one of the shells (Fig. 8).

When microscopically analysed, eight of the whole Theodoxus shells and the cervid tooth revealed traces of suspension on some kind of element that, after some time, due to movement, caused some areas of the holes to be polished (Álvarez Fernández, 2006, Dupont, 2006, Stiner, 2003, Stiner et al., 2013 and Vanhaeren et al., 2005). The location of these traces on the region closer to the natural hole of the shell specimens reveal that they were hanging with the heaviest part of the shell facing down (Fig. 9).

These characteristics can also be found in one of the Trivia shells that revealed the presence of what seems to be two holes on both extremities of the top surface, mind the fact that this is partially absent. The analysis revealed two polished areas (Fig. 10). This suggests that the shell was used as an adornment before it was broken. Finally, the cervid tooth also reveals traces of use on both sides of the bi-conical perforation. The friction between the tooth and a suspension element polished both the inside and the border of the hole causing its wear (Fig. 11). Regarding the clay ring fragment (Fig. 12), its whole surface, including the interior of the hole, exhibits several dark dots consistent with fire exposure, which might have been produced during the manufacture of the artefact or after, while exposed to some sort of fire source. The semi-circular hole has no traces of being polished and the absence of the other half makes it difficult to realise if the ring was used as a personal adornment before it was broken.

There is no microscopic evidence of other surface alterations other than those reported above for the studied sample.