6.0. Artifact analysis.

This chapter is an analysis of the Kom el-Hisn artifacts -- ceramics, lithics, floral and faunal remains, and epigraphic material -- in terms of their overall type distributions and how they are distributed among different kinds of deposits. Their spatial distributions are examined in Chapter 7.

6.1 Ceramics.

6.1.1 General description of ceramic types.

All ceramics in each excavation unit were saved. Body sherds were separated from diagnostic elements and sorted according to fabric type (Nile A, B, and C and marl), bagged, weighed, and stored on-site. Diagnostic sherds were washed and sorted according to a typology developed specifically for Kom el-Hisn. The typology is somewhat paradigmatic and is based on overall morphology, fabric type, and surface treatment. See Appendix II for definitions and drawings of a sample of each type used in this analysis.

A total of 27 types were defined over three seasons. Each successively defined type was numbered sequentially with odd numbers (1, 3, 5, etc.). Some of these were subdivided into additional subtypes and lettered sequentially (3A, 3B, 3C, etc.). These subdivisions were based on several factors including rim angle and shape, decoration and rim form.

These types correspond in large part to the whole-vessel categories derived from the standard corpus of Old Kingdom pottery types initially developed by Reisner (1931, 1932, 1942, 1955; also Mond and Meyers 1937; Brunton 1927). I have, therefore, assigned each Kom el-Hisn type to a corresponding standard Old Kingdom type as derived from mortuary assemblages from the Giza plateau and Saqqara. Reisner's original typology, while derived from purely mortuary assemblages, has continued to be used in settlement contexts with some modifications. It appears that, due to standardized mass manufacture of ceramics (probably in only a few production centers), there is a widespread correspondence between ceramics found in tomb and settlement contexts.

It should be noted, however, that both Reisner (1955:67) and Giddy (1987) are of the opinion that many cemetery vessels were produced specifically for that context. Giddy, in an initial comparison of the settlement and mortuary ceramics in the Old Kingdom areas of Dahkla oasis, suggests that the cemetery pottery "[appears] to have been fabricated solely for use in a burial context, and is consequently of much poorer quality and finish than the pottery from the settlement site" (1987:230, footnote #194). Reisner (1955) also notes that all or most of the ceramics from the tomb of Hetep-heres were unused and made (bought?) especially for burial. It is unclear, however, whether wholly different vessel types were produced for mortuary contexts (though some highly specialized forms do exist) or if the same forms were produced somewhat differently for mortuary and settlement uses. In some cases -- for example the offering jar (type IV in the Mycerinus typology) -- vessels produced for mortuary/temple use were subsequently recycled and used in everyday contexts. Reisner also describes one tomb vessel that had a manufacturing flaw rendering it otherwise unusable. Consequently, until extensive descriptions of Old Kingdom settlement assemblages are published the correlation between settlement and mortuary vessel forms should be treated as tentative.

As with most generalized typologies, Reisner's type definitions contain a mix of functional, technological, and stylistic criteria, though it appears to be predominantly functional in nature. Types were generally defined on the basis of overall form (especially the type of base), kind of clay and quality of manufacture, and surface treatment. Decorative elements tend to differentiate subtypes rather than defining entire types. The attention to function is understandable since there is a great deal of textual and pictorial evidence indicating the kinds of uses certain vessels were put. In addition, many vessels were placed in the tombs in their use context for use in the afterlife, often containing the materials (or residues thereof) that they were meant to hold.

Stylistically, there is little in the typology itself to indicate chronological variation for two reasons. First, Old Kingdom ceramics were mass-produced, probably at only a few centers, and were extremely utilitarian in nature. Compared with earlier Predynastic pottery, Old Kingdom forms lack the variety in shape and decoration characteristic of earlier periods when local production dominated. Second, from a typological viewpoint, Old Kingdom types were defined in a mortuary context where chronological concerns were not of paramount importance when defining the types since other dating methods (i.e., textual references) were more readily available. Some studies have suggested that chronologically significant variation does exist in various metric dimensions of vessel form (e.g., Ballet 1993). The practical relevance of this is that stylistic variation is only available through a detailed attribute analysis and is beyond the scope of this work.

Since the Kom el-Hisn typology is based on the standard Old Kingdom corpus, the types in this analysis are also based on overall morphology, fabric, and surface treatment. Of the 27 types defined thus far, only those types that had profile drawings and could with reasonable confidence be placed within a whole-vessel category were used in this analysis. An initial set of morphological types were created that combined the types and subtypes into larger categories of similar vessels corresponding to Reisner's types. A total of 12 morphological types were created from 19 of the ceramic types.

6.1.2 Ceramic type descriptions.

The majority of Egyptian ceramics are made of Nile silts with varying amounts of temper. Nile silt is defined as clays that were deposited by the river between the upper Pleistocene and the present (Bourriau et al. 2000) and were obtained from the Nile floodplain or nearby canals. These silts are rich in organics, iron compounds, and silica content, are brown to black when wet, and fire to brown or red in an oxidizing atmosphere. These have been classified into three categories based on the size and quantities of inclusions present, mostly sand and straw (from Boodle in Mond and Meyers 1937 p. 188). These are:

An additional source of clay are the marls originating in the shales and limestones bordering the Nile between Esna and Cairo with secondary deposits such as those from the Wadi Qena (Bourriau et al. 2000:121). Lucas and Harris (1962) describe the dominant sources of clays from outcrops near Qena and Ballas in upper Egypt (see also Nordstrom and Borriau 1993). Marl clays contain little to no organic matter but are high in calcium carbonates. They are generally brown or gray when wet and fire to pink or gray-green when fired. The marls have been divided into three types, A, B, and C, with four variants in the A-type defined by the size and type of inclusions. All have some form of limestone inclusions reflective of their source. Very few marl ceramics have been found at Kom el-Hisn and all of the diagnostic elements are of First Intermediate period dates (ca. 2134-2040 BC).

The following is a brief description of each Kom el-Hisn type with its corresponding type(s) from the standard corpus where applicable. Table 6.1 displays the metric data associated with each type. The data for Table 6.1 was derived from profile drawings for typed sherds. Only those drawings with diameter and fabric type were used. The average thickness was obtained by measuring each sherd profile in three places (fewer depending on the size of the sherd), calculating the mean thickness of each sherd from these measurements, and then calculating the mean for each type from these. To increase sample size, all sherds with good data were used, even if they were not from excavation units used in this study.

Type A: Bowl with tubular spout (Appendix II, Figure II.1).
KeH types: 7A,B,C

Corresponds to Reisner's types D-XXXV and D-XXXVI. Vessels of this type are often shown in tomb paintings depicting the beer brewing process and thus are interpreted as used for filtering beer, the spout facilitating the draining of the liquid portion into smaller vessels. They have rolled or recurved rims, and may have either flat or rounded bases. Those found in mortuary contexts most often have rounded bases but Reisner (1955) does not believe the shape of the base represented an essential functional difference. The Kom el-Hisn vessels are generally fine-textured of Nile B or C ware. Subtype is determined by the length of the spout.

Type B: Jars with rounded or pointed base (Appendix II, Figure II.2 and Figure II.3).
KeH types: 1B, 9A,B,C

Corresponds to Reisner's Type A-II (probably A-IIb). These functioned as generalized containers of liquids, most often represented pictorially as wine or beer jars. Reisner (1955:69) notes that they were particularly abundant in 4^th Dynasty graves and were often found closed with mud or plaster stoppers. Most of the Kom el-Hisn examples are B ware of moderate hardness. Many were described as cooking jars, though only 9 of the 45 drawn profiles have evidence of burning indicated. A red slip was noted on most samples. The subtypes of Type 9 are based on the amount of curvature just below the rim. Examples of subtypes 9A, B, and C are shown in Figure II.2 with a reconstructed vessel in Figure II.3. Subtype 1B represents a base (no examples shown).

Type C: Ordinary traditional offering jar (Appendix II, Figures II.4, II.5, II.6, II.7, II.8, and II.9).
KeH types: 1A, 11A,B,C,D

Corresponds to Reisner's type A-IV. This vessel apparently originated as a beer jar during the 3^rd Dynasty (Reisner 1955:70) and was commonly used as grave equipment during the Old Kingdom. They seem to have been specially produced for funerary uses (for both interment and as offering jars), but were often recycled as general-use vessels for carrying a variety of materials, including water and plaster (Bourriau indicates their use for 'every conceivable purpose'; 1981:61). While similar in overall shape and size to the pointed jars of Type B, these were of much coarser ware and were handmade by either coiling or low-rotation wheel, and they generally had a wider and less constricted opening than the Type B vessels. Several examples are shown in Figures II.4-6. Bases for these vessels (subtype 1A) vary from sharply pointed (Figure II.7) to somewhat flat as indicated by the two reconstructed vessels shown in Figures II.8 and II.9. The subtypes of Type 11 are based on the amount of curvature just below the rim.

Type D: Bread platter (Appendix II, Figure II.10).
KeH types: 3A,B, 13.

Corresponds to Reisner's Type F-XXVI ("Offering trays"). These are very large, thick coarsely made vessels and have been found in some tomb contexts holding food offerings or other smaller vessels. They have also been associated with baking throughout Dynastic times (Arnold 1982; Aston 1996; Nagel 1938), though Samuel (2000) believes this connection is tenuous. They have been found in association with ovens (Peet 1921:177) and the 12^th Dynasty tomb of Intefiqer shows a platter being used for baking (Davies and Gardiner 1920:14, plates 8, 9, and 9A). In addition, the fabric, construction, and wall thickness are all very similar to definite bread molds (Type E).

Many of these were drawn but, due to the size and shape of the vessel, diameters could be calculated from only 6 (see Figure II.10 for two examples). Type 13 tended to be thinner and the rim more flared than those of Type 3. Type 13 may also have had a more concave base. The manufacture of these vessels seems to be related to the bread molds (Type E) in that they are both coarse, soft Nile C wares of similar color and texture, and handmade. Many also have red slipping on the interior (exposed) surface.

Type E: Bread mold (Appendix II, Figure II.11).
KeH types: 3D,E,F

Corresponds to Reisner's Type F-XXV. Used to bake bread (a 12^th Dynasty tomb model shows a woman tending a stack of the Middle Kingdom forms over an open fire), this is one of the more ubiquitous ceramics found at Kom el-Hisn and, as Reisner (1955:88) notes, are common in graves from the 3^rd Dynasty through the 6^th Dynasty. Similar forms have been described at Qau and Badari (Brunton 1927) and Dahkla (Giddy 1987) where they dated to Dynasties 4-6. They are crudely made vessels thought to be produced by pressing clay over a wood or stone form (the interior surface is very smooth) and many found at Giza had designs imprinted on the interior surface to show on the finished loaf. The clay is usually a soft Nile C or any readily available clay. Their generally cheap construction is probably a function of both their ubiquity and the fact that they were often broken to extract the bread (Bourriau 1981:65). Only 15 of the 91 drawn specimens show evidence of slipping: 11 are slipped on both interior and exterior surfaces, 3 on the exterior side only, and 1 on the interior side only.

Type F: Flared bowl (Appendix II, Figures II.12 and II.13).
KeH types: 17, 19, 35A, 49A,B

Corresponds to Reisner's Type D-XXXIX (Flaring flat-bottomed bowl). This type represents a somewhat variable set of flared bowls. The clay type is primarily Nile B with a few of Nile C and most often are medium to very hard. Types 17 and 19 are heavier, coarser, and larger vessels than Types 35A and 49A and B, but are of similar overall shape. Type 17 contains a rope impression on the exterior just below the rim. Type 19 is a subtype resembling Reisner's Type D-XXXIXc in which a ridge is molded into the interior of the vessel just below the rim. Whether this ridge is functional or stylistic is unclear; Reisner suggests it may be derived from a particular type of stone bowl with a cup hollow.

Type G: Bent-sided bowls (Appendix II, Figures II.14, II.15, and II.16).
KeH types: 23AB, 27AB, 35B, 37AB

Corresponds to Reisner's Type C-XXX (Bent-sided bowls and basins). These vessels are especially abundant in 5^th and 6^th Dynasty tombs. The majority of the Kom el-Hisn specimens are finely made, most of a hard Nile B clay and most (130 of 165 drawn specimens, or 79%) were slipped. Reisner interpreted the bent shape and fine, hard fabric as indicative of water or other liquid storage/serving rather than as than cooking vessels. Smaller vessels of this type have been described as drinking cups (Bourriau 1981:19,61). Type 23 is from a larger vessel of more variable size, probably of the 'basin' type, and of somewhat cruder manufacture, some of which are of clay type C. Type 35B tends to be smaller than the rest with less variation in size and of relatively finer construction.

Type H: Meidum (carinated) bowls (Appendix II, Figure II.17).
KeH types 31A-G, 33ABCD

Corresponds to Reisner's Type C-XXXII. This is a common form found all over Old Kingdom Egypt. It is a finely made ware of very hard Nile B or A clay, and nearly all have some form of slip applied. There is a great deal of variation in the degree of carination present (which may have chronological significance; Ballet 1987) as evidenced by the number of subtypes associated with these vessels. They are often represented as holding food or liquid and occasionally as a decorative dish holding flowers floating in water (Bourriau 1981:53). At Kom el-Hisn, preliminary analyses suggest that they tend to be associated with faunal remains and features associated with food production and consumption (Wenke et al. 1988a:27).

Type I: Round-bottom bowl with molded rim (Appendix II, Figures II.18 and II.19).
KeH types: 29A-F.

Corresponds to Reisner's type C-LXIII. The type is similar in most respects to other round-bottomed bowls, but differentiated by the molded rim with an interior groove just below the rim. The bowls themselves have a concave to slightly flaring wall in some cases, and are predominantly of a medium-hard Nile B clay.

Type J: Little brown bowls (Appendix II, Figure II.20).
KeH types: 39

Type 39 is the typical small brown bowl with a slightly beveled rim that is characteristic of the 12^th Dynasty. Various uses are applied to this type but they are generally regarded as drinking cups as they are often placed beside the deceased's mouth as burial equipment. Some specimens (e.g., #91 in Bourriau 1981: 67) have a pinched rim and functioned as oil lamps. At Kom el-Hisn most (65%) had some form of slip applied and the majority were of Nile B clay type and very hard. The color of the slip ranges from brown (10YR5.3) to light red (2.5YR6/6). These were abundant in the Middle Kingdom areas of the site, but were also found in small quantities in other parts of the site, presumably due to either sediment mixing or misidentification with a bowl of similar type, O (see below).

Type K: Basin with roll rim (Appendix II, Figures II.21 and II.22).
KeH types: 25

Corresponds to Reisner's Type C-XXIX. Some of these may be parts of Type A vessels since the shape of the rim in both are nearly identical. Reisner also states that the two are very similar with this type having a round bottom and no spout (1955:79). Fabric types are evenly distributed between Nile B and C and all are medium-hard. The majority (74%) have some slipping evident.

Type L: Flat bowls or plates with contracting mouths (Appendix II, Figure II.23).
KeH types: 61

Corresponds to Reisner's Type D-LXXII. These vessels may have been used to contain some sort of liquid since there are a few examples from Giza that had small spouts (subtypes D-LXXIIc and d) and the contracting mouth also indicates the containment of liquid. All of the drawn examples are of Nile B ware of medium hardness. Another vessel form, goblets, have similar rim shape and diameter; but the distinctive bases of these types have yet to be found at Kom el-Hisn.

Type M: Conical bread molds (no drawings).
KeH types: 2A-G

No corresponding Old Kingdom type. These are typical Middle Kingdom conical bread molds. They closely resemble their Old Kingdom counterparts in fabric, manufacture (coarse, soft Nile C ware) and function.

Type N: White bottles (no drawings).
KeH types: 43ABC

No corresponding Old Kingdom type. These sherds resemble those from a form of First Intermediate Period (2134-2040 BC) bottle (e.g. #21 on page 2 of Bourriau 1981). They are made of marl clay, hard-fired and have a white slip.

Type O: Round-bottom bowls (Appendix II, Figures II.24 and II.25)
KeH type 41

Corresponds to Reisner's type C-LXI (Round-bottom bowls with plain rim) and possibly C-LXII (Round-bottom bowls with contracting rim). These are similar to the Little Brown Bowls of Middle Kingdom date (Type J) but tend to be larger (p=0.000 for 2-tailed t-test comparing rim diameters) and more red than brown in color with a less tapered/beveled rim. Functionally, they were probably equivalent to the Middle Kingdom variety (e.g., drinking cups, infant feeding cups, and lamps, the latter two by pinching in a portion of the lip; Bourriau 1981:67) and are abundant at Kom el-Hisn. Fabric is predominantly a medium-hard Nile B. Of the 114 specimens 76 have some form of slip.

6.1.3 Chronological issues.

Several ceramic types present have chronological significance within the period encompassed by these occupations. The majority of the deposits at Kom el-Hisn (and the majority of those excavated) are of Old Kingdom date. However, a few areas contain deposits of First Intermediate or Middle Kingdom date. The two ceramic types present at Kom el-Hisn that are typically Middle Kingdom are Type J (Little Brown Bowls) and Type M (Conical bread molds). The Middle Kingdom deposits are restricted to the southeastern portion of the excavated area near unit 1261/1074 where a 12^th Dynasty scarab seal was found in DU-6 (SU-14) below the adjacent brick wall in unit 1263/1074. An analysis of the distributions of Type M ceramics shows that this type is most abundant in the area surrounding 1261/1074 and decreasing in frequency away from this unit. While unit 1261/1074 itself contains only one Type M sherd, a nearby unit, 1256/1072 (excavated in 1988 and not used in this analysis) contained abundant Type M sherds (153 of 194 total sherds from SU-1, or 79% of the total). The only other unit with a large number of Type M sherds is 1219/1095 (22 of 193 or 11.4% of the total). The Type M sherds found outside of the explicitly Middle Kingdom units are found exclusively in either the Upper Pottery Layer or dump deposits close to the surface.

Type J vessels, Little Brown Bowls, are also indicative of Middle Kingdom dates, but the distribution of this type is somewhat more problematic. Generally, they are common in those units defined by other criteria as Middle Kingdom. For example, in 1261/1074 Type J sherds make up 70% of the identified sherds (109 of 156). In a nearby unit 1256/1072 (not used in this analysis) they make up 19% of the total identified sherds (37 of 194). Unexpectedly, a significant number were also found in 1219/1095 which was ostensibly of Old Kingdom age. The Type J sherds found in this unit occurred in virtually all deposits throughout the sequence, notably in the intact floor deposit of DU-7 and the wall collapse of DU-6, both deposit types that are tightly associated with the existing architecture.

Apparent Type J sherds also are found in small amounts in widely scattered locations all over the site. The numbers tend to be small, usually between one and four sherds in an entire excavation unit/room; those units with a relatively large number (1192/1035 with 17 and 1235/1056 with 14) have them concentrated in UPL, dump, and other reworked deposits. Unit 1192/1035 is thought to be composed of a variety of sloping dump or sheetwash deposits which may also explain the wide variety of sherds found in this unit. This would indicate that their distribution was due to transport and mixing with earlier deposits, but several Type J sherds are found in otherwise sealed Old Kingdom deposits in floors and wall collapse deposits.

I suspect that these "Type J" sherds are probably of a related type, O (Kom el-Hisn Type 41), which is very similar in overall shape to Type J. As noted in the type descriptions above, Type O bowls differ from Type J as being larger and more often red than brown in color and have a more rounded rather than beveled rim. In terms of spatial distribution, Type O sherds are more widespread, occur more often in strictly Old Kingdom contexts and occur less often in otherwise Middle Kingdom deposits than Type J sherds. Furthermore, Type O is more abundant in later levels (1 and 3) than in earlier (4) levels. It is therefore possible that, due to the similarity in the two kinds of sherds, that some misidentification could have occurred.

Type N, Small White Bottles, are of marl clay and are thought to be First Intermediate Period and later (Wenke et al. 1988a:27; Bourriau 1981). Only 36 diagnostic sherds were found in widely dispersed locations across the site and are most often found in dump and Upper Pottery Layer deposits, though some are found in wall collapse deposits. Unit 1192/1035 contained several (8) indicative of this unit's formation from transported dump and sheetwash deposits. Several were also found in the apparent Middle Kingdom deposits of units 1261/1074 and 1219/1095 though these were all in UPL, or dump deposits so they may be there because of dispersive processes as is assumed with the other locations. Because of the unpatterned distribution of these sherds they are not thought to convey much information of chronological significance as far as the sequence of occupation is concerned.

6.1.4 Ceramic distribution among deposit types.

The distribution of ceramic types among different deposits can provide some indication of their use environment. Schiffer (1987), for example, has related the rate of waste production (discard) to the manufacture rate multiplied by a waste/defect production constant. This latter term may also be expressed as the failure rate during either production or use. Assuming broken items are not left in occupied areas, those ceramic types with higher discard rates should have correspondingly higher manufacture rates and/or failure rates. Consequently, those types that are more prone to manufacturing or use breakage should be discarded more often and thus found more often in dump deposits.

To determine whether any types occur more often in certain deposit types than others, I employed Chi-square tables crossing the presence/absence of each type with the deposit types. To ensure some minimal level of reliability to these tests I initially dropped fluvial, UPL, burials, and column bases from Old Kingdom levels since they are rare and result in numerous cells with expected values <5 and minimum expected values of <1.0. After inspecting the resulting distributions, I also dropped pits because several of these (e.g., the clay linings) can be expected to have few or no ceramics at all and were causing most of the observed significant results. That is, most of the significant probabilities were being driven primarily by the relative absence of ceramics from the pit deposits.

When pits are dropped from the analysis (leaving only dump, floor, and wall-derived deposits), only types C (p=0.003), F (p=0.037), K (p=0.003), and to a lesser extent E (p=0.078) show significant patterning when all levels are combined (a =0.05). When only Level 3 deposits are used only types C (p=0.003) and K (p=0.000) show significant patterns (a =0.05). Levels 4+, 1, and 2 have too few deposits for reliable results. Residuals indicate that all of these are found significantly more often in Dumps and less often in Floors (Type C is also found significantly less often in Wall deposits as well). Some factor or factors are causing these vessel types to be discarded more often than other vessel types.

Types C (Ordinary traditional offering jars) and E (Bread molds) are large, heavy, poorly made vessels that are cheap to manufacture and consequently easy to replace. The presumed functional regime of these two types (cooking) would also subject them to thermal shock on a regular basis resulting in more opportunity for mechanical failure. Another poorly made vessel, Type D (Bread platter), was generally not used in a heated environment and consequently not subjected to as many opportunities for failure.

The other two types, F (Flared bowls) and K (Basins with roll rim), are more problematic. Neither is particularly abundant relative to other types. Unlike Types C and E, these are made mostly of the harder Nile B clays (87% and 73%, respectively) and although the available data is not sufficient to determine the firing regime, it is undoubtedly higher than the very low temperature firings of Types C and E. The ratio of rim diameter to wall thickness for these two types is also close to those of other bowls (0.04; see Table 6.1) suggesting that they may not be inherently more fragile structurally and therefore more prone to breakage and discard. However, they both have the largest diameters of all types which, combined with their presumed function of water/liquid storage, may render them more prone to breakage apart from the simple rim diameter:thickness ratio. Reisner (1955: 65) describes most of this type as having a black fracture profile which may indicate a lower firing temperature and thus a softer and weaker final form contributing to more breaking. I suspect that their increased presence in dump deposits is probably a combination of a somewhat softer fabric on a large vessel and their probable use in beer brewing -- a common staple of the Egyptian diet -- which would lead to increased failure. Their large size would make them relatively expensive to manufacture which might explain the overall low frequency of this type throughout the site. A more detailed analysis of the performance characteristics of these vessels may elucidate this issue further.

I next conducted a one-way analysis of variance to determine if any ceramic types occur in any deposits with greater frequency than expected. I used only the three main deposit types (Dump, Floor, and Wall) used in the Chi-square analysis above to avoid spurious results from rare deposits with few ceramics. I also used the logarithmically transformed densities to satisfy the normality and equal variances assumptions. The results largely confirmed those obtained in the Chi-square analysis. Types C, E, and K were all found in significantly higher frequencies in Dumps than in Floors or Walls (p=0.002, p=0.001, and p=0.023, respectively). In addition, Type H (Meidum bowls) were also found in greater abundance in Dumps than in Floors or Walls (p=0.006). Type G occurred in roughly equal frequency in Floors and Dumps, both higher than in Walls.

Among the other types, G and H are only found in statistically higher densities in Dumps as opposed to wall-derived deposits; the densities of these types between Dumps and Floors are not statistically different. I do not deem this particularly significant since those sherds appearing in wall-derived deposits are there only incidentally or through their use as chinking material which is not a primary function of the vessels themselves. Thus, the only types that seem to be found significantly more often in dumps and in higher frequencies are Types C, E, and K. I have argued above that Types C and E are cheap to manufacture; Type C vessels are used in a variety of contexts; Type E (bread molds) are used in an environment conducive to thermal shock and resulting breakage, and textual sources also indicate these molds were often purposefully broken to remove the bread; Type K may be structurally weak because of a relatively soft fabric and use as a container of large amounts of liquid. Thus, their abundance in dump deposits could be their simple low-cost, abundance, and use in environments that promote failure. However, it is also possible that the relative absence of these types from floor deposits results from their not having been used in this area of the site, yet discarded here. This will be examined in the next chapter when spatial distributions are examined in more detail.

6.2 Stone tools

Lithic analyses were performed by M. Kobusiewicz. The lithics are divided into three broad categories: chipped stone tools, debitage, and stone objects (Table 6.2). The latter category contains ground stone tools and any other non-naturally occurring rock not contained within the other two categories (basically, all non-flint objects). The chipped stone tools and debitage are grouped according to the basic stone tool typology derived from Tixier (1963, 1974).

6.2.1 Chipped stone tools.

A total of 367 pieces of chipped stone were recovered from the 1986 and 1988 excavations. These are broken down as shown in Table 6.3. Some examples are shown in Appendix II Figure II.28a-e and IIf-g.

The raw material for both debitage and retouched tools is predominantly Egyptian flint (N=360, 98.09%), a fine-grained brown-to-gray chert common throughout the desert margins of the Nile valley. A coarser chert made up a small number of specimens (N=5, 1.36%) and an additional 2 specimens were unidentifiable due to extensive burning. The source of the flint is outside of the Delta and the Nile floodplain proper. Weathered flint nodules may be found in extensive Plio-Pleistocene gravels up and down the valley and along the margins of the Delta. In situ nodular and tabular flint is also found in outcrops of Eocene-Oligocene limestones. The most extensive outcrops of this material are found to the south near Bir Kiseiba but it is also available nearer the Delta at Abu Roash (Aston et al. 2000).

Cores. Only three cores were found during these two seasons, two of flint and one of a coarser chert (two were also found in 1984). One was a single platform core (flint) in the early stages of preparation (Figure II.28g). The platform was prepared by striking a few small flakes from the sides of the nodule and slightly roughening the surface. Only one full flake appears to have been struck from this core. The other two cores (one of flint the other of coarse chert) were circular or discoidal in shape with prepared platforms (Figure II.28f). Neither seems to have been used for the production of blades. All three had cortex present in either primary (>50% covered) or secondary (<50% covered) coverage.

Debitage. The 232 pieces of debitage (minus cores) are broken down in Table 6.4.

Of those with discernible platforms (N=93) pointed platforms dominate with N=80 or 86.02%. The remainder are distributed between lisse (N=5, 5.38%), faceted (N=3, 3.23%), cortex (N=3, 3.23%), and other unidentifiable platform types (N=2, 2.15%). 77 (96.25%) of the 80 unretouched blades with a discernible platform are of the pointed type, the remaining being faceted and lisse. Similarly, pieces with no trace of cortex also prevail with 197 or 84.91% of the total; of the blades only, 191 of 197 or 96.95% have no cortex. Of the remainder, 6.84% (N=16) have secondary cortex present, and only 1.72% (N=4) have primary cortex present.

Most of the blades (N=184 of 197) are broken, probably intentionally to eventually fit them to sickle handles. Of the 13 whole blades, the mean length is 75.85 mm (+ 12.73), the mean width is 14.38 mm (+ 3.28), and the mean thickness is 4.38 mm (+ 1.39). Length ranges from a minimum of 56 mm to a maximum of 95 mm; width ranges from 10 to 19 mm, and thickness from 2 to 6 mm.

Some form of excessive heating or burning is evident in 50 (21.34%) of the debitage specimens, 36 of which (72.0%) are blades. Of the 50 burned specimens, 34 (68.0%) are in deposits that do NOT show other evidence of burning; only 13 (36.11%) of the 36 blades are in burned deposits. A chi-square test reveals that the distribution of between burned and non-burned deposits is statistically random. That is, burned pieces of debitage are not found preferentially in otherwise burned deposits, implying that there is no necessary connection between the burning evident in certain deposits and the burned or unburned state of these lithics. Still, of the cores collected over all

three seasons, only two show any evidence of heating and both of those were described as only slightly burned or fire-cracked. Consequently, I interpret the heating evident on some lithic specimens to be most probably the result of post-manufacturing accidental heating rather than heat treatment for any technological or functional purpose.

Retouched tools. The most abundant form of retouched tool is the sickle blade which comprises almost 85% of the total in Table 6.5. Examples are shown in Appendix II Figure II.29. The majority of sickle blades are broken, which Kobusiewicz interpreted as a result of manufacturing -- purposely breaking them to size for fitting into a sickle handle -- rather than from use. In addition, fully 83.72% of the sickle blades are central fragments (both distal and proximal ends missing), while 11.63% are proximal ends, and only 4.65% are distal ends -- also indicative of purposeful breaking. Average dimensions for all sickle blades are: Length: 33.09 mm (+ 13.51); Width: 12.95 mm (+ 3.10); Thickness: 3.85 mm (+ 1.40). Since the average length of sickle blades is less than half that of unretouched blades (and assuming that the unretouched blades are the source of the sickle blades), it is apparent that the unretouched blades were broken further before fitting into the final tool. This indicates that blade blanks rather than finished sickle blades were being brought to the site where final manufacture and fitting took place, as well as (presumably) maintenance.

As with the debitage, only 29 (26.36%) sickle blades have evidence of burning (recalling that 18.27% of blade blanks had evidence of burning). And again, there is no apparent correlation between their location in burned or unburned deposits.

The majority of sickle blades (69; 62.72%) have retouch only along one edge, split roughly equally between the left and right sides. The remainder (N=41; 37.27%) have retouch along both edges. Somewhat surprisingly, only 54 (40.60%) have polish present from use; the presence of polish is also about evenly divided between those with one and two retouched edges.

Of the remaining retouched tools, all of the bifaces are broken (Figure II.28a-e). Eight are of unknown type, however one was probably a bifacial knife which was subsequently broken and then reworked along one edge for other purposes. Of the four fragments of bifacial knives, three are central fragments and one is probably part of the tang. The smaller of the two endscrapers has a concave working edge and the larger (Figure II.27e) has a convex working edge (both are whole).

6.2.2 Ground stone tools.

The distribution of ground stone tools is shown in Table 6.6. The majority of ground stone objects are made of a fine sandstone or quartzitic sandstone (N=252, 63.16%), followed by limestone (N=70, 17.54%), marl (N=53, 13.28%) and other materials (N=24, 6.02%) which includes chert, alabaster, conglomerate, and others. The distribution of raw materials among flakes and grinding stones is very similar (88.42% and 92.68%, respectively) suggesting that the flakes were derived from modifications to the grinding stones.

Of the grinding stones, most (87.80%) were fragmentary. Of those that were identifiable to upper (mano) or lower (metate), the vast majority (N=48, 90.57%) were from metates. Five of the metates were whole, the remainder, fragmentary. The largest of these is from unit 1192/1035 DU-2 and measures 260X132 mm. A smaller one from Room 12 DU-1 is made of limestone and measures only 25X13 mm, and was probably used for grinding something other than grain.

Only a single whole mano was found, from Room 13 DU-1 and measured 75X63 mm. At least one of the fragmentary specimens gives a good idea of the overall shape and size of the original, that from Room 17 DU-1, which measures 81X100 mm and was split longitudinally and has one end broken off (Appendix II, Figure II.26).

The chunks present a mixed bag of material and object characteristics. Many are probably fragments of grinding stones without any worked surfaces present. A sizable portion are of limestone and marl, however, of which there is little represented in the grinding stone assemblage.

Of the remaining objects, several merit some mention. An elongated piece with grooves cut or worn into both ends appears to be a net weight (Appendix II, Figure II.27). This was found in Room 23 along with a whole sandstone pounder, possibly a small grinding stone. Room 23 also contained a metate fragment and several sandstone and limestone fragments, suggesting some of the manufacturing may have been done within that room. Several (8) globular chunks of chert were also found which do not appear to have been cores for chipped stone tool production; half of these were found within a single room, 5, suggesting their use as tools of some sort. Finally, a fragment of a polished alabaster palette was found in unit 1219/1095.

6.2.3 Summary of stone tool assemblage.

The majority of chipped stone tool production involved the specialized manufacture of sickle blades. Most were struck from single platform cores using a soft hammer technique and at least one edge was retouched for use. Because the majority of sickle blades had only one edge that was used, they were probably hafted and used only until dull and then discarded.

Little evidence involving the manufacture of any chipped stone tools is available apart from the tools themselves. Due to the very small number of cores and core trimming elements, the lack of debitage characteristic of bifacial reduction, and the overall lack of cortex present on any specimens, one must conclude that lithic production was either carried out off-site or in restricted areas within the site which have yet to be excavated. Even a detailed microscopic analysis of sediment samples (Cagle 1991) failed to produce more than a few isolated flakes that could represent the remains of trimming or retouch. This lack of manufacturing debris or even debris associated with tool maintenance leads me to believe that most of the chipped stone tools were imported in finished, or nearly finished state. This would explain the general single-use nature of the dominant lithic type, sickle blades. However, the large number of unretouched blades suggests that at least some retouching, fitting, and maintenance took place on-site. The raw material would have had to be imported, possibly from as near as Abu Roash near Giza or as far away as southern Egypt; it seems unlikely that transport of whole, unmodified cores would have taken place over such long distances.

A different situation exists for ground stone tools. Much evidence of reshaping in the form of small flakes exists to support the hypothesis that a good deal of modification occurred within the site, possibly at the location of use. Large scale production is not in evidence, but maintenance and remanufacture into other items seems to be somewhat common. The most common stones -- sandstone, limestone, and marl -- are all abundant in northern Egypt; harder granites that are only located far to the south near Aswan are very rare at Kom el-Hisn.

All of the tool forms and distributions at Kom el-Hisn are well known from other sites (Ginter, et al. 1979; Midant-Reynes 1983, 1985; Schmidt 1992a, 1992b) reflective of an agricultural economy. At least as far as the stone tool assemblage is concerned, Kom el-Hisn conflicts with the idea that rural Old Kingdom settlements were relatively self-sufficient in terms of commodity production. The evidence presented indicates that most stone tools were imported in finished or nearly finished form with minimal modification thereafter at least for chipped stone tools.

6.2.4 Stone tool distribution among deposit types.

Table 6.7 presents the mean densities for each class of stone object for each of the major deposit types sorted in descending order of density. The high densities of debitage and ground stone for pit deposits is largely due to a single outlier, Room 5 DU-9 which is a very small deposit (.01 m²) and contains a single piece of debitage and two ground stone objects. Without this deposit, the density for debitage in Pits is 0.618 + 1.52 and that for ground stone would be 2.02 + 4.27, putting both just above Fluvial deposits.

UPL deposits have the highest concentrations of stone objects of all types, owing to its nature as a lag deposit. Among the other deposit types, wall-derived deposits, as might be expected, have the lowest densities of stone objects. Pits (without Room 5 DU-9) also have few lithics overall, though they tend to be quite variable. Again, this is due to the mixed nature of the deposits themselves: those that might otherwise be classified as dumps (e.g., Room 5 DU-9) have higher densities than others, such as those composed of clay linings (e.g., Room 2 DU-7 and 8) or collapsed bricks from the pit structure itself (e.g., Room 13 DU-3). Dumps have higher densities of debitage and ground stone than Floors, but Floors have higher densities of retouched tools than Dumps (for all levels and Old Kingdom levels only). None of these are statistically significant.

This pattern is consistent among individual types of objects as well with a few exceptions. For example, there are more ground stone flakes in floors than in dumps. This is not statistically significant either but it does suggest that the inhabitants were maintaining or modifying their ground stone implements themselves within the habitation structures. The most abundant retouched tool, sickle blades, also are found in greater densities in floors than dumps (again, not significantly so). Beyond this, there is little observable patterning. Floors tend to have higher densities of burned and polished sickle blades, and also higher densities of those with retouch on one side only, but not significantly so. This latter pattern suggests that sickles were more readily disposed of after both sides were used.

Overall, deposits with high densities of debitage do not often have high densities of retouched tools. However, there is a linear correlation between sickle blade density and unretouched blade density (r=0.575, p=0.000). This correlation only holds within floors and walls; dumps do not show any correlation and generally have higher densities of unretouched blades than sickles. This is consistent with the interpretation that the residents were finishing the sickle blades themselves at least partly within the habitations.

There does not appear to be any significant patterning in ground stone object types or raw materials across deposit types. The only materials with abundance measures amenable to statistical testing are sandstone, limestone, and marl. Sandstone and limestone are not more or less abundant in any deposit type, though limestone does have a higher mean density in Dumps. Marls are marginally (p=0.088) more abundant in Dumps than Floors.

The preceding observations largely support the conclusions reached from the analysis of the tool types alone. Very little chipped stone debitage is present and that which is found is usually blade blanks presumably to be used in the production of sickle blades. At least some sickle blades were finished and maintained on-site, as indicated by the correlation between blade blanks and retouched sickle blades in floor deposits. Sickle blades were often used along only one side and then discarded, but there is some evidence for greater discard to dumps of sickles with retouch on both sides. Floors also tend to have greater concentrations of finished tools, with dumps containing somewhat greater concentrations of debitage suggesting the removal of manufacturing waste from occupation areas (Clark 1984; Gallagher 1977; Schiffer 1987).

Ground stone tools show somewhat different distributions. Ground stone objects as a whole are generally found in greater concentrations in dumps, except for flakes which are found in greater densities in floors. There is no apparent relationship between the occurrence of ground stone objects and manufacturing/maintenance debris (chunks and flakes) in any deposit type, and ground stone implements are not more abundant in floors than dumps. A plausible interpretation of this pattern is that of the inhabitants maintaining or modifying their ground stone implements in their places of apparent use (floors) and discarding those that are beyond repair, which would result in more amorphous chunks and broken objects in dumps. However, one would expect that this kind of manufacturing debris would, like their chipped stone counterparts, be removed to refuse areas; such is not the case. It is possible that many of these flakes may have been put to other uses while the irregular chunks and shatter were removed to refuse dumps, but analyses of possible wear on these objects have not been performed.

The overall impression one gains from this analysis is that most stone implements were imported in largely finished condition from outside of Kom el-Hisn, with some finishing and maintenance performed by the inhabitants. Chipped stone tools were probably imported as either prepared cores or more probably blade blanks and then retouched as needed. The abundant flakes and chunks from ground stone implements suggest that a good deal of maintenance or reshaping for other uses was performed at the site itself, apparently in the rooms where the objects were used. This suggests a degree of dependence on some basic goods and services that is at odds with the traditional impression of villages and towns being largely independent and self-sustaining.

6.3 Faunal remains.

Bone and shell were removed from the screens (or during excavation), dried in the sun, washed, and brushed of excess dirt before being analyzed by Richard Redding in the field laboratory (Redding ND). The material was initially sorted into shellfish, fish, reptile, bird, mammal, and unidentified groups. Shellfish remains were not further identified but only weighed and counted. Fish remains were identified to at least genus level; that which could not be identified to genus was sorted by element: cranial, vertebrae, post-cranial, and unidentified. These were both weighed and counted in 1986 and 1988. Reptiles and birds were identified to the lowest level possible, weighed and counted (bird material was identified by Steven Goodman). Mammal bone, where possible, was identified to at least genus level, counted and weighed. The remainder was sorted into Large and Medium skull, vertebrae, and rib. Teeth were also identified as either bovid, pig, canid, or other. All of these were counted and weighed.

6.3.1 Fish.

A total of 1750 fish bones were recovered from the 1986 and 1988 seasons. Of these, 364 could be identified to at least genus level. Table 6.8a presents the genus breakdown of the fish remains. The distribution of body parts for each species is presented in more detail by Redding (ND) for the 1984, 1986, and 1988 seasons combined, but individual distributions are not available for the set of deposits used here. Consequently, further references to body-part composition of individual species are taken from Redding's analysis of the three combined seasons' assemblages.

Fish are considered to have been an important subsistence item during dynastic times and before. Fish remains are common in both tombs and settlement sites and are commonly represented pictorially and in texts. They were captured by a variety of methods, including hooked lines, nets, spears and, in certain situations, simply gathered from shallow water (Darby et al. 1977). Tomb representations indicate that fish were most often dried, either with or without vertebral columns and/or heads. There is little information available as to the specific processing methods performed on individual species, which makes interpretations based on species representation problematic.

The most common type of fish found at Kom el-Hisn is Tilapia, a type of chiclid, two species of which are found in the Delta region. They inhabit shallow, near-shore water, and their meat is considered to be very good. Cranial elements represented approximately 58% of the total, a figure much higher than that at Merimde (von den Driesch and Boessneck 1985:95). Redding (ND) suggests that the proportion of cranial elements at Merimde is what one would expect if whole fish were entering the archaeological record; that is, they were being caught and consumed locally. The higher proportion of cranial elements at Kom el-Hisn is thus thought to reflect preferential processing of fish rather than consumption whereby the heads were removed and the body shipped elsewhere. This pattern may also reflect processing in this particular area of Kom el-Hisn with consumption taking place at another location.

The second most abundant fish is Synodontis, an omnivorous bottom feeder that prefers deep, open water. Synodontis is considered edible, but rather dry and flavorless. The majority of specimens (88.7%) are cranial elements. Analysis of the proportions of individual elements indicates that pectoral girdle and dermatocranial elements are actually underrepresented at Kom el-Hisn suggesting that these fish were imported into the site with their heads already removed. The proportions of body parts are also comparable to those at Merimde (von den Driesch 1986; von den Driesch and Boessneck 1985) indicating processing was similar at both sites.

Sparus auratus and Bagrus sp. are found in approximately equal numbers, 58 and 57 specimens, respectively. Sparus auratus is a marine fish whose meat is considered excellent and was undoubtedly imported into Kom el-Hisn. As at other sites (Tell Maskhuta, Minshat Abu Omar and Tell El Dab'a, though not at Merimde; von den Driesch 1986), this genus is represented only by pharyngeal plates and teeth. Bagrus is found in the Delta and is a carnivorous bottom feeder. Bagrus bayad was considered sacred. The Osiris myth relates that when Osiris was cut into pieces by his evil brother Set the phallus was thrown into the Nile and was consumed by three types of fish one of which Plutarch called 'Phagrus'. Possibly because of this cultic association thousands of Bagrus were mummified (Brier 1994; Brier and Bennett 1979) and placed in tomb contexts. The majority (87.1%) are cranial elements, though as Redding (ND) notes, the post-cranial skeleton is soft and does not preserve well, and pectoral spines are considered as cranial elements in his analysis. Still, the foregoing suggests that these two types of fish were brought into the site whole.

Clarias (Nile catfish), of which two species are found in Egypt (C. angularis and C. gariepinus), prefers shallow, poorly oxygenated water and can be caught easily during the breeding season in shallow pools. The meat is considered edible but not particularly good. About 2/3 of the specimens are cranial elements.

The remaining fish are found in small quantities. Tetradon fahaka is a puffer fish reported occasionally from other sites. It is found at Kom el-Hisn as jaw elements and the occasional tooth and is considered edible. Lates niloticus, the Nile perch, is considered to have excellent meat, but is represented at Kom el-Hisn by only six specimens, all from the 1984 season and all cranial.

Table 6.8b compares the rank orders of fish remains (based on NISP) from several sites. As Redding (ND) indicates, the distribution of fish remains at Kom el-Hisn most closely resembles that of the Neolithic site of Merimde and the Early Dynastic Minshat Abu Omar, both Delta sites. The major differences between the Kom el-Hisn and Merimde assemblages are: Clarias sp. is extremely abundant at Merimde and comparatively rare at Kom el-Hisn; Sparus auratus is abundant at Kom el-

Hisn and rare at Merimde; and Merimde has a larger number of rare species. The greater species richness at Merimde may be a function of much larger sample sizes there. Minshat Abu Omar has more abundant Tilapia sp. and Lates niloticus than Kom el-Hisn, while Bagrus is absent at Minshat Abu Omar.

Of the four most common types at Kom el-Hisn, Tilapia sp., Synodontis sp., Sparus auratus, and Bagrus sp., Sparus and Synodontis both seem to have been imported from some distance away (probably Bagrus as well) with only the most abundant Tilapia providing some evidence of local capture. The higher relative abundance of these species at Kom el-Hisn may reflect the more developed trade networks in place in the Old Kingdom. This suggests that there was at most a portion of the community or perhaps a nearby satellite community engaged in fishing for locally available species. The similarities with Minshat Abu Omar and Merimde further imply that similar divisions of labor hold there as well. Aldred (1987:58) notes that fish were considered ritually impure foods and were generally consumed by the working classes, particularly artisans working on state projects who were given a daily ration of fish along with their other provisions.

6.3.2 Birds and reptiles.

Few reptiles remains have been found over three seasons of excavation. A single snake vertebrae, probably from the family Colubridae, was found in the upper pottery layer of unit 1235/1056. Several species of this family are found in the Delta region and they were probably not consumed at Kom el-Hisn. A single lizard bone was found in 1984, but could not be identified to either genus or family.

A total of 62 turtle bones were recovered, but the majority came from a single room, 12. The specimens in Room 12 probably came from a single animal, Trionyx triunguis, the soft-shelled turtle. The carapace from this animal was found largely intact in one corner of Room 12 and was probably butchered within the confines of this room. Difficult to catch, the meat from this species is considered excellent. The remaining turtle bones are of unknown species, but may be either Trionyx or Testudo graeca, a tortoise also considered to have excellent meat.

A variety of bird remains have been found at Kom el-Hisn owing to its position in the Delta, a prime wintering area for many Eurasian species. Tomb representations and textual references indicate that birds were widely hunted and consumed in dynastic times. 200 bone fragments were identified as avian; identification for the 1988 season is not available. Of the 178 avian bones from the 1984 and 1986 seasons, 23 were identifiable to species (plus 9 specimens identified by eggshell only). The species composition is as follows, in order of abundance:

Anas sp. (duck): N=12;

Struthio camelus (ostrich): N=9 (all eggshell fragments);

Fulica atra (coot): N=6;

Ardea cinerea (Gray heron): N=1;

Platalea leucorodia (Spoonbill): N=1;

Gallinula chloropus (Moorhen): N=1;

Porphyrio madagascariensis (Purple Gallinule): N=1

Tringa cf. erythropus (Red shank): N=1

Ostrich (Struthio camelus) is generally identified by eggshell fragments and only rarely by bones. They are thought to have entered Egypt as tribute from Nubia and used as food, containers, and for bead production.

All of the identified species are found to winter in Egypt from about mid-September through April, and all are found represented on tomb paintings in various fowling, offering, and procession scenes and their remains are found occasionally as food offerings in tombs. Some (e.g., the purple gallinule, Gallinula chloropus) are also sold for food in the modern Delta region. All of the above (except ostrich) are water birds and are non-nesting inhabitants of the Delta during the winter months. All of the bones examined have fused epiphyses indicating none were young enough to have been hatched during this period.

Compared to other sites, Kom el-Hisn is most similar in the dominant species of ducks (Anas) and coot (Fulica atra) to Giza (Kokabi 1980; Redding 1992) and Tell el-Dab'a. At Merimde the dominant genii are ducks (Anas), geese, and rail, while coots are represented only by a single specimen. The dominant species at Giza (gray cranes) and Tell el-Dab'a (flamingoes) are entirely absent from Kom el-Hisn. In the latter case this may be environmental since flamingoes are abundant near the coast where Tell el-Dab'a is situated. This suggests that much of the fowling was done locally by either inhabitants of the sites themselves or by those of nearby settlements.

6.3.3 Mammals.

Mammal bone makes up the bulk of animal remains found. A total of 12,126 identified bone fragments were recovered from the 1986 and 1988 seasons. Of these, 866 could be identified to at least the genus level (Table 6.9); the remainder was identified to body part (Table 6.10).

Bos taurus: Only 11 bones could be identified as belonging to domesticated cattle (compared with 20 total for the 1984 and 1986 seasons). The ratio of cattle to sheep/goats at Kom el-Hisn is 0.06:1 and the ratio of cattle to pigs is 0.04:1. These are generally much lower than at other sites. Giza has a cattle:sheep/goat ratio of 0.8:1 and a cattle:pig ratio of 43.7:1 (Kokabi 1980:520; Redding 1992); at Merimde the cattle:sheep/goat ratio is 0.6:1 and the cattle:pig ratio is 0.5:1 (von den Driesch and Boessneck 1985:6); the Predynastic areas of Hierakonpolis have a cattle:sheep/goat ratio of 0.8:1 and a cattle:pig ratio of 5.7:1 (McArdle 1982:16). Data supplied by Mark Lehner from the workman's village at Giza show the cattle:sheep/goat ratio is 3.1:1 and the cattle:pig ratio is 32.7:1 (Redding 1992:105).

This suggests that cattle were much less of a subsistence item than were pigs or sheep/goats. This does not necessarily negate the hypothesis that Kom el-Hisn could have functioned as a specialized cattle-raising station and other data indicate that cattle were common in and around the site. The plant data analyzed by Moens and Wetterstrom (1988) suggest to them that much of the plant material found at Kom el-Hisn came from cattle dung used as cooking or other fuel. The character of the plant material further suggests that the cattle were kept close to the site for at least part of the year and fed cut fodder (see section 6.3.4 below for a more detailed discussion of these data). Additionally, of the nine Bos bones for which fusion data was possible, only one was from an animal that lived past the age of two years (Redding ND:19). Thus, it appears that only young animals were being processed for food, probably those that died of natural causes. At other sites where cattle are relatively more abundant, the number of adult animals is much higher. For example, in eight out of nine excavation units at Merimde, adults outnumber juveniles by at least 3:1 (von den Driesch and Boessneck 1985:7) and at Hierakonpolis mature cattle account for 75% of the samples (McArdle 1982:117).

Redding (1992, ND) has developed a model of cattle production that may account for the apparent lack of cattle in the Kom el-Hisn assemblages. In this model, the cattle produced at Kom el-Hisn would have been exported to other locations, presumably for elite consumption or as provisions on state projects. Local consumption would be restricted to the very young that died before reaching an appropriate age for export, and the very old. If, as Redding suggests, the preferred animals for export were males between 1.5 and 2 years of age, then one would expect that the destination sites would be dominated by male animals in the 1.5-2-year old age range. The Old Kingdom workmen’s village at Giza is one such state project, and in that assemblage (based on fusion data from first and second phalanges) between 84% (second phalange) and 88% (first phalange) of the animals were less than two years old (Redding 1992:105). Of the five fragments from which sex could be determined, four were males. Thus, at least one site that could have received estate-raised cattle conforms to this model.

It seems unlikely that cattle, an abundant part of the faunal assemblages from a wide variety of sites from different time periods in Egyptian history, were entirely absent at Kom el-Hisn. The possibility remains, however, that the processing and consumption of cattle remains are in as yet unexcavated portions of the site. It is also possible that cattle remains were disposed of differently from other types of remains. C.C. Edgar, for example, noted that an area to the north end of the site contained an abundance of what appeared to be cattle bones (Edgar 1919-1915:63). Consequently, while a great deal of circumstantial evidence points to the presence of cattle despite their absence archaeologically, more direct evidence must await further archaeological and perhaps textual research.

Ovis/Capra: Sheep and goats represent 315 specimens. These animals served a variety of purposes during Dynastic times as sources of meat, milk, leather, and fiber (wool). The ratio of meat-bearing to non-meat-bearing bones is similar to that of complete animals (Redding 1992, ND) and, since most of the remains come from floor and dump deposits, it is likely that the animals were being slaughtered and butchered within or close to the areas of the site that were excavated.

The kill pattern for the animals for which age data is available (N=82) indicates that 75.6% were slaughtered after they reached two years of age. This is similar to the 70% at Giza, but higher than that at Merimde and Hierakonpolis where the percentage of adult animals is 50% or less. The Merimde and Hierakonpolis figures correspond to a pattern that maximizes the amount of meat for local consumption. The Giza and Kom el-Hisn figures may have several sources. The Kom el-Hisn kill-off pattern is similar to that constructed by (Payne 1973) for maximization of wool production. It is also possible that younger animals were shipped off to urban areas for consumption there. The sex ratio of three females to every one male at Kom el-Hisn (Redding 1991, 1992, ND) also supports the idea of male animals being preferentially removed for consumption elsewhere.

Sus scrofa: Pigs are the dominant remains at Kom el-Hisn. 422 specimens were identified as pig (2860.1 g), representing almost half of the total identifiable remains recovered. Pigs were primarily used for food in dynastic Egypt and the Kom el-Hisn data corresponds to such a strategy. The ratio of meat-bearing to non-meat-bearing elements is 0.55:1 compared to 0.37:1 in a whole animal (Redding ND:26) and the number of cranial fragments recovered indicates that, along with sheep and goats, the animals were being slaughtered and butchered in or near to the area of excavation. Almost half of the pigs for which age data is available (N=76) were between the ages of one and two. At Merimde, the only other site where kill-off data is available (based on tooth eruption), the majority of pigs were slaughtered before the age of six months (von den Driesch and Boessneck 1985:25). Both Harris (1985) and Redding (1991) have argued that there is an inverse relationship between the intensive growing of grain and the incidence of pigs. Pigs will only be abundant in an agricultural settlement when there are sufficient food sources besides grain, when raising grain is a secondary activity, or when there is sufficient surplus available for grain to be fed to pigs. Kom el-Hisn may meet the first two requirements for abundant pigs if the hypothesis of a specialized cattle-producing station is correct, and may be contrasted to Merimde where a more generalized agricultural community seems to have existed (that is, less specialized). In this scenario, pigs could be kept at Kom el-Hisn for consumption by the local population while not competing for the same food resources.

The remaining domestic taxa consist of animals for which no clear evidence of subsistence use exists. Equus asinus, the domestic donkey, has 37 specimens, 30 of which come from a single excavation unit (1261/1074, which is not of Old Kingdom date). The domestic dog, Canis familiaris, is represented by 26 specimens, although the jackal (C. aureus) may also be part of the sample. The cat (Felis sp.) is represented by only two elements of unknown species.

Several wild taxa are present that demonstrate that some hunting of wild game was still in effect during the Old Kingdom, though this is not thought to represent a significant part of the diet. A gazelle of unknown species is represented by three specimens and an addax (Addax nasomaculatus) was recovered from the 1984 season. Both of these species were probably hunted in the western desert.

31 specimens of Bubal Hartebeest (Alcelaphus buselaphus) were recovered (the 482 bovid teeth in Table 6.10 are thought to be from this species as well). These animals are common at Egyptian sites and were probably hunted in the desert areas bordering the Nile valley. The ratio of meat to non-meat bearing elements is 0.37:1, compared to 0.47:1 for a whole animal, suggesting that the animals were brought to the site intact.

A pair of hippopotamus bones were also recovered from Room 17; element composition is unknown. Room 17 also contained several hartebeest and gazelle remains.

Rodent specimens numbered 17 in all and are of unknown species. However, Redding (ND) reports that at least one mandible belonging to the house shrew (Suncus murinus) was identified in the 1986 excavations. This shrew is not native to Egypt and probably entered the country (probably as a stowaway) as part of a trade shipment(s) from its native Indian subcontinent. A cape hare (Lepus capensis) was also located in the 1984 samples.

6.3.4 Distribution of fauna among deposit types.

As noted in Chapter 4, dumps contain higher densities of faunal material than do any other deposit. In terms of NISP density (NISP/m³) dumps have significantly higher densities of total identified specimens than other deposits, and significantly higher densities of unidentified bone than other deposits.

The densities within individual groups of species (cattle, sheep/goat, pigs, fish, birds, and rodents) are also consistently higher in dumps than in other deposits, with one exception: three pits, Room 5 DU-8, 9, and 11, all contain high densities of fish bone (40 g/m³, 500 g/m³, and 44 g/m³, respectively). The number of bones in each is not particularly excessive (NISP=3, 7, and 2, respectively), but they are all relatively small deposits. Perhaps not coincidentally, Cagle (1991) also noted the presence of numerous fish cranial elements and teeth in the sediment sample for Room 5 DU-9, as well as in a similar pit structure in Room 2 (DU-6). The material used in that study is from flotation samples and much of the recovered bone would have passed through the mesh used for the bulk of the sediments used in the present analysis. Because of this, one might posit that these pit structures were often used to dispose of fish remains after the primary function of the pits had been completed. It also suggests an area where more detailed examination of fish remains is needed.

The distribution of unidentified faunal remains is also consistent with the inhabitants regularly removing animal remains from occupied areas. Dumps contain far higher densities (by weight) of these remains than other deposits, as one would expect were the inhabitants regularly removing food remains to refuse deposits.

The presence of burning in the deposits does not seem to have a significant impact on the amount of bone present. Dumps with burning evident had a higher average density of both identified and unidentified bone than unburned dumps, but not significantly so. Conversely, floors with no burning evident actually had higher densities of identified and unidentified bone than did burned floors but again not significantly so. Burned wall deposits, however, had a significantly higher average density of both identified and unidentified bone than unburned walls (p=0.000). The only observable pattern that might explain this is that most of the burned wall deposits (12 of 13) were deposits where the bricks or brick fragments were visible and had not decomposed in situ or been redeposited. I would argue that these deposits were probably formed by active burning and perhaps demolition of the structure with some accompanying dumping of refuse.

Other patterns emerge when individual taxa are examined in burned and unburned deposits. Ovicaprids and pigs in particular have markedly different distributions. There is little difference in density when all deposits are considered together. However, pigs are more abundant in burned (9.30/m³ + 11.83) than unburned (5.99/m³ + 5.08) dumps, but are less abundant in burned (3.29/m³ + 5.26) than unburned (8.50/m³ + 15.04) floors (the differences are not significant at the 0.05 level). Birds follow this same pattern, though there is little difference in densities between burned and unburned Floors. Ovicaprids, however, are more abundant in unburned (5.90m³ + 7.40) than burned (2.99/m³ + 3.28) dumps and also more abundant in unburned (3.08/m³ + 8.73) than burned (1.52/m³ + 3.01) floors (neither is significant at the 0.05 level). Fish also are more abundant in unburned dumps but there is little difference between burned and unburned Floors.

While the relatively high standard deviations makes interpretation necessarily tentative, if one assumes that burned deposits are the refuse from cooking, the relatively low levels of ovicaprids in burned deposits suggests that they were not often consumed in this part of the site. The pattern of pig remains suggests there was some butchering done in the occupation areas along with cooking and discard of the remains. This coincides with Redding's interpretation of ovicaprids as primarily used for purposes other than consumption while pigs were raised, butchered, and eaten locally.

Among fish species, there is generally less patterning evident. Only four species are abundant enough to meaningfully compare densities: Tilapia sp., Synodontis sp., Sparus auratus, and Bagrus sp. Both Tilapia and Synodontis are relatively more abundant in unburned than burned Dumps (not significant at the 0.05 level). The remaining species are approximately equally distributed between burned and unburned deposits.

Species composition is largely consistent across deposit types: pigs are dominant, followed by fish, ovicaprids, and birds, in dump, floor, and wall deposits. One exception to this pattern is pit deposits where fish remains comprise the vast majority of faunal remains, with low levels of pigs and ovicaprids, and virtually nothing else. In addition, rodent remains were low in most deposits, but substantially higher in wall deposits. Wall deposits contained significantly higher densities of rodent bones than floors or dumps. Rodents also were absent from redeposited collapse and intact walls and occurred exclusively in wall collapse and decomposed wall collapse. I interpret this to be a function of the density of the sediments, wall collapses being less dense than intact walls or redeposited material, the latter of which is sorted into relatively finer, and therefore more closely packed, sediments.

In sum, the distribution of faunal remains among different deposit types is what one would expect in a situation where the inhabitants were actively removing food debris from occupied areas and transporting them to refuse dumps. With the exception of pits, the species composition within deposit types mirrored the overall abundance of taxa in the site as a whole. The inhabitants also apparently had some predilection for disposing of fish remains in pit structures.

The pattern of abundances in burned or unburned deposits seems to support the interpretation that sheep and goats were kept primarily for their milk and wool with some consumption of adult animals, while pigs were raised locally primarily for their meat. Butchering and cooking seem to have taken place in the same general area.

6.4 Floral remains.

A total of 82 flotation samples are used in this analysis, distributed among 57 DU's; of these 48 were Old Kingdom. The species identification was carried out by two different investigators and in some cases the species identification was not consistent between the 1986 and 1988 seasons (W. Wetterstrom analyzed the 1984 and 1986 floral remains, A.C. D'Andrea analyzed the 1988 remains). Consequently, I have chosen to use only those species that were identified in both the 1986 and 1988 seasons. Because of this, some of the species distributions will differ somewhat from the published data (e.g., Moens and Wetterstrom 1988).

Recovery techniques also differed somewhat between the 1986 and 1988 seasons. In 1984 and 1986 a single bucket flotation system was used in which the flotation sample (usually 3-4 liters in volume) was poured into the bucket containing water and the floating plant remains were decanted into a sieve covered with fine mesh cloth. In 1988 a double-bucket apparatus was used in which water was agitated up through the interior bucket containing the sediment with the floated material passing through a spout and into a fine mesh sieve. In both cases, flotation samples were air dried before processing and the light fractions (the floated plant material) further air dried before analysis. Both sets of remains were analyzed using binocular microscopes (10X to 50X for the 1988 material, unknown magnification for the 1984 and 1986 material).

6.4.1 General description of plant remains.

Totals for each DU sampled are provided in Appendix I, Tables I.5a, I.5.b, I.5.c, I.5.d; summaries for individual taxa are provided in Table 6.11. A total of 10,719 specimens are reported, of which 3,593 have been identified to at least the family level or below. By and large, the percentages of identified specimens correspond with those given by Moens and Wetterstrom, with one major exception: reeds and sedges make up 22.5% of their total identified specimens while those in Table 6.11 make up only 5.16%. This results from an identification anomaly: the 1988 data do not contain either Phragmites australis or Carex sp. (a reed and sedge, respectively) which together make up a large portion of the 1984 total.

The cereal remains and weeds are probably both the result of grain processing. The majority of the chaff is composed of glume bases and spikelet forks, mostly of emmer wheat, which are separated during threshing and winnowing of the grain. The weed species are also probably a result of grain processing. Darnel grows almost exclusively in cultivated fields and canary grass is common around modern fields. Both of these have been found in tombs as contaminants of grain offerings (Schweinfurth in Schafer 1908:154-157).

Reeds and sedges are common in the marshy environment of the Delta and probably entered the site incorporated in the dung of cattle though they also could have been burned directly as fuel. The higher percentage of wild grasses in the 1988 samples may explain the anomalous absence of grasses in the 1984 samples reported by Moens and Wetterstrom: reeds and sedges would be expected to be present along with wild grasses if cattle were foraging in the immediate environs of Kom el-Hisn, since grasses remain common even among these other marsh-dwelling plants.

Among the fodder plants, the clover identified at Kom el-Hisn is probably berseem (Trifolium alexandriunum), one of the principal crops grown in modern Egypt, and is considered an excellent fodder material. The seeds in the archaeological samples are smaller than their modern counterparts; Moens and Wetterstrom suggest that the Kom el-Hisn seeds may have shrunk during carbonization or possibly represent an older variety of berseem.

Little charcoal has been recovered, suggesting dung and/or plant stems rather than wood as the principal fuel. Dung fuel is usually prepared by mixing the dung with straw or some other tempering material, forming it into cakes, and allowing it to dry. Such methods are known from modern Egypt and throughout Pharaonic times. The chaff from grain processing could have served as a temper in this context and winnowing debris is also known as a major animal feed (van Zeist and Camparie 1984:4-5). Very few sheep/goat pellets have been found which tend to be common when used as fuel (Bottema 1984:208). Cattle are thus the most likely source for the Trifolium since cattle are often represented being hand-fed cereal, cereal stalks, straw, and cut fodder (Moens and Wetterstrom 1988:170) while pigs and sheep/goats were usually allowed to forage on their own.

The large amount of fodder plants indicates that cattle were fed for at least part of the year with cut fodder, presumably while they were kept in pens within or near the site. This may have occurred during the inundation when dry areas of the Delta were few and far between, generally on levees, along the desert margins, and (probably not coincidentally) on gezira deposits, upon which Kom el-Hisn rests. During the remainder of the year (approximately 9-10 months) the cattle may have been herded to more distant locales for foraging of wild grasses, explaining the relative lack of wild plant material at the site (though some local grazing probably did occur). It is also possible that the animals were kept in pens during a portion of the year and fattened with high-quality fodder in preparation for sacrificial offerings. In fact, two classes of cattle are referred to in representations and texts: iw3-mdt or stable cattle are generally shown being fed cereal, cereal stalks, or cut fodder (e.g., Montet 1954: 43-58; Lepsius 1849-1858, Pl. 62) while ng3-cattle were depicted as being thinner, used for agricultural labor, and herded into fields for grazing (e.g., Montet 1954:74).

6.4.2 Distribution of plant remains among deposits.

Density values in this section are based on the number of identified specimens per unit weight of sediment in the flotation sample. As one might expect, plant density is higher in those deposits that have evidence of burning, though a T-test only gives a probability of 0.075, just missing the cutoff for significance at the 0.05 level. Table 6.12 provides descriptive statistics for each class of plants. Field weeds and fodder plants have the highest mean density, followed by chaff, other identified plants (primarily grasses), cereal grain, and reeds and sedges. Contra D'Andrea (1989) and Moens and Wetterstrom (1988) I found that species diversity is not particularly evenly spread among deposits. Species richness is tightly correlated with the number of identified specimens (p=0.000) and the density of NISP (p=0.000); however, there is no relationship between either NISP or NISP density and the amount of sediment in the sample. Thus, one can interpret the apparent relationship between NISP and species richness as a function of the depositional process rather than the sampling regime.

As noted above, the density of plant remains is largely determined by the visible presence of burning in the deposit. The source of burned material is probably cooking fires, though other forms (such as trash burning) cannot be ruled out. Those floors with the highest densities of plant remains tend to be those where discrete areas with high concentrations of burned material were noted; e.g., Room 18 DU-8 and Room 1 DU-3. Probably not coincidentally, these two DU's also have the highest density of cereal remains of all deposits, save for Room 17 DU-2 which is wall collapse immediately adjacent to an obvious hearth (Room 17 DU-3). Those floor deposits with more diffuse areas of burning or no burning at all (e.g., Room 9 DU-7) have few plant remains.

To further examine this pattern, Figure 6.1 shows the mean densities for each category of plant by deposit type (the sole Burial, UPL, and Fluvial samples excluded). Floors tend to have a more even distribution of plant types than other deposits (save pits, which have little plant material overall), while dumps have a more uneven distribution, dominated by weeds, chaff, and fodder. Walls (primarily Wall Collapse deposits) have a distribution quite similar to floors. I suspect this is largely due to some mixing of floor deposits with overlying wall collapse, though the inclusion of hearth debris in the wall material itself cannot be ruled out (the only intact wall sample, 1235/1056 DU-20, has very few plant remains).

The plants represented in dumps are what one would expect if winnowing debris and spent fuel from cooking fires were being discarded. The more even distribution of remains in Floors suggests that a variety of activities -- e.g., winnowing, cooking -- were taking place within the sampled structures. Indeed, as already mentioned, those Table 6.12 deposits with discrete areas of burning have higher concentrations of cereal grains than others. Further patterning (such as different areas for different activities) is not readily apparent from the plant data alone.

The following general relationships seem to hold: 1) Those deposits with high densities of plant remains (NISP density) tend to have high species diversity as well; 2) Dumps have a more uneven distribution of plant types than other deposits, possibly indicating the remains of winnowing debris and spent fuel; and 3) Floors have more even distributions of plant types, with cereal grains being more abundant in areas where discrete burning is evident.

A provisional interpretation of these trends is that the excavated area contains some rooms where food preparation and cooking were performed, accounting for the abundant cereal grains, with chaff and fodder material that was probably incorporated in dung fuel. Adjacent rooms, such as Room 9, contains little or no plant material and probably have little to do with food preparation. Dumps may have as their source the cooking and winnowing debris from nearby rooms, as the dumps sampled are high in plants one would expect would have been incorporated into dung fuel and also direct winnowing remains.