Ceramic Standardization and Intensity of Production: Quantifying Degrees of Specialization

by Valentine Roux
Ceramic Standardization and Intensity of Production: Quantifying Degrees of Specialization
Valentine Roux
American Antiquity
Start Page: 
End Page: 
Select license: 
Select License



Valentine Roux

The ainz qf this study is to e.xamine the relationship behceen lnetric variabiliv of ceramic vessels of a single type und interz- sity ofproduction. This relationship is e.iar?zined on the basis of 11essels made by Indian and Spanish potters whose rates of production van from low to high. Reslilts are compared with Filipino data to ernbmce different cultures and therefore pos- sibly different ernic conceptions of standardizution. We concllide that rate ofproduction affects the degree of sta~zdardi:a- tion. However, only in a high-mte prod~~ctio~z

situation do w3e have motor habits that transcend emic conceptions of smndardi:ation. Indeed, factor> like size classes, emic conceptiotl of size classes, and skill may induce a certain variabil- ih despite similar rates ofproductio~z. Indexing degrees of smndardizatior1 needs, in other respects, to take into account the ciinzillarive effect ofprodlrction events tlzatplay on the CV (Coefficieizt of Variation) values. Applied to Mesopotamian data, our resu1t.r suggest that third r7~illenni~irn

ceramic production can be compared to lo>v-mte production obsewed erhno- gmphically.

El propdsito de este est~rdio e.r ana1i:ar la relacidn entre la variubilidud mitrica de Ins vasijas de un solo tipo cerdmico y la intensidad de la produccidn. Estu relacidn se estlrdia a partir de unn serie de piezas manllfacturadas por alfarervs de In India y de Es1mAc1 clrya produccidn puecle vnriar en cnntide de baja a alta. Los resultados se comparan con datos de 1os.filipinos paru aburcnr diferentes c~rlt~tms

Jpor lo tunto, posibles diferencias en conceptos imicos de esmndarizacidn. Concluirlzos que la tcua de produccidn incide en el nivel de estandarizncicin. Sin embargo, solamente en caso de una tasa altu de produccicitz hay lzabitos de manufacturcl qlre trunscienden 10s conceptos emicos de estarzdari~acidn, En efecto, factores como Ins clases de dimensiones, 10s conceptos Pmicos del tamaAo y la destresa del urtesano producen cierta variabilidad a pesar de qlre huyn tasas sinlilares de produccidtz. Pam medir 10s gmdos de estandarizacicin se requiere, entre otros aspectos, tomnr en c~rentu el efecto acumulativo de 10s eventos de produccid~z sobre 10s vnlores del CV (coejcienre de variacidn). A1 aplicar estos resitl- tados a 10s datos de Mesopotcimicl, .se cuentn con bases para plnntear qLte la produccicin de ceranzica en el tercer milenio puede ser corrlpurnda con linn produccidn de r~ivel bajo tal conzo se lo ha observado en contexros etnograjcos.

n archaeology, the degree of ceramic stan- Sinopoli 1988; Stark 1995). Degree of standard- dardization is taken to reflect intensity of pro- ization may be assessed through raw material com- duction and degree of specialization. The position, manufacturing techniques, form and standardization hypothesis (e.g., Costin 199 1, dimensions, and surface decoration. Two types of 2000; Costin and Hagstrum 1995) proposes that attributes may be distinguished (Costin and more uniformity is due to a higher rate of produc- Hagstrum 1995): intentional and mechanical. tion. The latter is correlated with economic spe- Intentional attributes are controlled by the artisan cialization, which encompasses many ways to and include technological, morphological, and styl- organize craft production (domestic vs. specialized istic properties. Mechanical attributes are those production, part-time vs. full-time potters, attached "which the potter unintentionally introduces into vs. independent potters, individual vs. workshop, his or her works" (Costin and Hagstrum 1995:622). etc.) (e.g., Arnold and Nieves 1992; Blackman et They relate to motor habits and skill, and, because al. 1993; Costin 1991,2000; Costin and Hagstrum they are unconscious, they more directly reflect the 1995; Feinman et al. 1981; London 1991; Lon- organization of production. In this regard, metric gacre 1991; Longacre et al. 1988; Rice 1989,1991; variability should reflect the number of work units,

Valentine Roux .Laboratoire de PrChistoire et Technologie, CNRS, 92023 Nanterre cedex, France

American Antiquity, 68(4), 2003, pp. 768-782 
Copyright0 2003 by the Society for American Archaeology 


"on the assumption that the amount of variability in these mechanical attributes correlated directly with the number of independent potters or work groups" (Costin and Hagstrum 1995:622). For a large quantity of pottery, the less the variability, the fewer the number of full-time potters.

The inverse correlation between metric vari- ability and rate of production has been partly tested by ethnoarchaeological studies (e.g., Kvamme et al. 1996; Longacre 1999). Statistical results have enabled the authors to differentiate, for aperture and height, between household potters and part-time andlor full-time potters, and for aperture between part-time and full-time potters. The authors con- clude that numerous factors (e.g.. differences in raw materials, potters' levels of expertise, market demand, manufacturing techniques, local tradi- tions, and types of measurement aids) affect ceramic standardization (see also Arnold and Nieves 1992), and that more effective means of measuring and assessing variability in ceramics would help "untangle relationships between pro- duction intensity and product standardization" (Kvamme et al. 1996: 125).

With this purpose in mind, we conducted an ethnoarchaeological study to isolate the parameters involved in standardization, as well as to quantify the relationship between intensity of production and degree of standardization. We examine vessels made by Indian and Spanish potters with different rates of production. We then compare results to Fil- ipino data to propose regularities (Gallay 1986) between rate of production and metric variability, taking into account different possible emic con- ceptions of standardization. Results are applied to archaeological data from Mesopotamia.

Ethnographic Case Studies

Three ethnographic situations are considered. Two of them are in India, in rural and urban contexts. They are characterized respectively by low- and high-rate ceramic production. These two situations, because they pertain to the same cultural context, will enable us to easily isolate the parameter "inten- sity of production" and assess its effects on stan- dardization. The third case, in Spain, is in an urban context. This enables us to compare high-rate ceramic production within two different cultural contexts. Moreover, the Spanish example enables us to compare high-rate production using different fashioning techniques: without rotative kinetic energy and with rotative kinetic energy (here, pad- dle-and-anvil and wheel throwing).

Low-Rate Ceramic Production (India)

Low-rate ceramic production was studied in the southern part of Andhra Pradesh (A.P.). Four vil- lages, next to Tirupati. were visited within the framework of an ethnoarchaeological survey1 (Chavaram Bekam, Alapakam, Kota Aruru and Chapakam). These villages are separated by only 3-10 km. Each of them contains 1,200-2,000 inhabitants and 2-5 potters. Potters work within the domestic space, in the courtyard or under an awning. There is no warehouse; pots are stored in the house. The firing area is generally located at the outskirts of the village. Potters work through- out the year according to different rhythms. Festi- val months like January and February represent the peak season, while monsoon months represent the slack season. In most cases, pottery is the only source of income of the potters. Pots are sold by potters' wives, either through patron-client links (the jajmani system) or through barter or trade, within a 34 km radius.

Three types of vessels represent the larger part of production and consumption: kura catfi. ralla catti and pedda bana.

Kura catti are vessels for cooking vegetables (for dishes like kura, samba, rasam). Ralla catti are vessels for cooking lentils (dal) and spinach that, once cooked, are ground against the bottom of the vessel with a wooden pestle. Pedda bana are big storage jars where mainly cereals are stored.

All in all, around 6,000 pots are produced each year per potter, out of which, roughly speaking, less than 50 percent are kura catti and ralla catti, and 5 percent are pedda bana. The other products are various cooking, transport, and storage pots of sev- eral sizes. The demand is not for strictly standard- ized vessels, but for a volume that may vary from one family to another depending on the size of the family.

Vessels are fashioned by a combination of wheel throwing and paddling. Pots are first thrown on a pivoted-spoked wheel. Depending on their dimen- sions, they are thrown off the hump (kura catti) or from a mass of clay whose quantity corresponds to the pot to be thrown (pedda bana). They are detached from the hump or from the wheel with a

AMERICAN ANTIQUITY [Vol. 68, No. 4, 20031

string in such a way that a hole is left at the bot- tom. Neck and shoulder dimensions do not usually change after this point. The lower part of the body is a thick wall, shaped by subsequent paddling; the paddling process starts when the clay has reached a leather-hard consistency, using stone anvils and wooden paddles. The first task is to close the hole by paddling the pot in a downward direction. Once the bottom is closed, ash is added to the inner lower part of the body that is shaped with a smaller anvil and a lighter paddle. For ralla catti, sand is added to the bottom as an abrasive against which vegeta- bles will be ground. Paddling follows. The big jars are paddled twice. Ash is added during the second paddling. Ash prevents the anvil from sticking to the clay.

The pots receive little decoration. While pots are still on the wheel, horizontal lines are incised with apiece of wood. Once they have dried, pots are cov- ered with a red slip.

High-Rate Ceramic Production

High-rate ceramic production was studied in North India and in Spain.

In North India, the study was conducted in Uttam Nagar, a suburb of New Delhi with about 200 potter households. Every house has a nuclear or extended family that owns one or two wheels, placed under open shelters, and one kiln located in the courtyard. The clay, conveyed by truck, is stored in the courtyard or in the street, depending on spa- tial constraints. Pottery manufacture is a full-time activity that takes place even during the monsoon; rains restrict the number of firings but do not stop production.

The different potters interviewed specialize in the manufacture of water jars (glzada or matka) made in three sizes: small (ghadiya), medium (bich ka ghada) and large (bara ghada). All in all, around 15,000 pots are produced each year per potter, of which, roughly speaking, 30 percent are ghadiya, 60 percent are bich ka glzada, and 10 percent are bara ghada.

Pots are fashioned according to the throwing and paddle-and-anvil technique described for Andhra Pradesh pots. However, as is common in Northwest India (Saraswati and Behura 1964),no hole is left at the bottom when the pot is detached from the wheel, which is a pivoted block wheel.

Spanish ceramic workshops have been studied by Arcellin-Pradelle and Laubenheimer (1982, 1985). These workshops are made up of special- ized areas, which include the manufacturing area, the drying area, the kiln, and the warehouse. In some cases, the house is part of the workshop area. Two to four potters can work together. Most assis- tants are not members of the household.

The production of one workshop has been exam- ined in detail and published extensively. It is located in Val1 de Uxo, a village in the province of Castel- lon. It is run by a single potter who carries out the whole manufacturing process. He works inten- sively (10 hours a day), 11 months a year. Vessels are sold mainly through middlemen. Production includes table and cooking vessels.

Pitchers with flat bottoms are considered here. They are wheel thrown on a kick-wheel. When the clay has reached a leather-hard consistency, the bottom is turned (scraped) on the wheel. The han- dle is fixed with slurry. The potter makes about 50 pitchers a day. Annual production is approximately 14,000 pitchers.

According to the craftsman, his intention is to make very similar pots; standardization, in his view, indexes the degree of skill.

Analytical Procedure

To study how rate of production affects metric vari- ability on series of vessels representing a single ceramic type, we examine series of vessels fash- ioned within low-rate vs. high-rate production con- texts. Series of vessels made by one potter is a production event and represent a "group." The hypothesis is as follows: if rate of production affects motor habits, then vessels made by high-rate pro- ducers will present less intra- and inter-group vari- ability than low-rate producers. Metric indices that may reveal motor habits remain to be defined on the basis of empirical data. A wide range of mea- sures is therefore considered.

Measurements of Andhra Pradesh ceramics include (Figure 1) height, maximum diameter, aperture (rim diameter), thickness of the wall and of the lip, and width of the lip. For the thickness, we took three measures on each side of the pot: two of them relate to the wheel-thrown part, one relates to the pad- dled part. Thickness and width of the lip were mea- sured on each side of the aperture. These measures





Maximum Diameter


Figure 1. Measurements taken on Andra Pradesh vessels (ralla cao.

enable us to calculate indexes of regularity in the fashioning of the walls and the lip.2

Measurement of New Delhi ceramics include height, maximum diameter and aperture (rim diam- eter).

For Spanish ceramics, measurements taken by Arcellin-Pradelle and Laubenheimer include (Fig- ure 2) (1)maximum diameter, (2) aperture, (3) neck diameter, (4) base diameter, (5) total height, (6) height of the maximum diameter, (7) lip height, (8) lip thickness, (9) handle width, (10) handle thick- ness, (11) height of the decoration (distance between the bottom and an incised groove), (12) height of the zone that has not been turned (scraped), (13) capacity, and (14) weight.


In Andhra Pradesh, dimensions were measured on 166 ralla catti, 186 kura catti and 85 pedda bana made by, respectively, 6, 7, and 6 potters.

Three potters (A, B, and G) live in Chavaram Bekam. They are respectively 45,35, and 32 years old. Two of them (A and B) are brothers-in-law. They live in the same compound.

Two potters (C and D) live in Alapakam. Son and father, they are respectively 30 and 60 years old, and live in the same compound.

One potter (E)lives in KotaAnuu. He is 45 years old. One potter (F) lives in Chapakarn. He is 65 years old. Except for two series of kura catti, the ceramic vessels were not fired, having been made the day

Figure 2. Measurements taken on Spanish vessels. (1) Maximum diameter, (2) aperture, (3) neck diameter, (4) base diameter, (5) total height, (6) maximum diameter height, (7) lip height, (8) lip thickness, (9) handle width,

(10) handle thickness, (11) height of the decor (distance between the bottom and an incised groove), and (12) height of the zone which has not been turned.

they were measured, or the day before. The fact that vessels were not fired and not measured the same day does not affect the results. Indeed, all the ves- sels were measured leather hard, that is to say when the drying shrinkage affecting the volume of a ceramic had already occurred. Moreover, the kura catti data did not reveal any significant difference of values between the fired and unfired series.

In New Delhi, measurements of 180 ghariya were made. These pots are fired and distributed by 6 potters who are, on average, 46 years old (the youngest is 30 and the oldest is 65).

In Spain, 100 fired pitchers were measured. These were made in the course of two days by one potter. Depending on the day of manufacture, they belong to series A or B. Series C includes series A and B.

Indexes of Standardization

Variability between low-rate and high-rate pro- duction is examined by comparing the inter-group variability presented by the two Indian assemblages (Andhra Pradesh vs.New Delhi). Let us recall that a group corresponds to apotter's production. Com- parison is justifiable since the fashioning process is the same in both situations.

Inter-group variability has been computed using ANOVA (analysis of variance). As underlined by Kvamrne et al. (1996), results obtained by an F-test are dependent on an assumption of normality that does not hold most of the time. For this reason,

AMERICAN ANTIQUITY [Vol. 68, No. 4, 20031

we proceeded with a posteriori tests (LSD3 [Scheffk] and Games-Howell) that are very robust with populations that are not normally distrib- uted and are useful for comparing heterogeneous variances on small samples.

Variability between low-rate and high-rate pro- duction is also examined by comparing the intra- group variability presented by the Indian (Andhra Pradesh and New Delhi) and the Spanish samples. The Spanish material will enable us to test whether the range of values, as observed in the Indian high- rate production context, varies according to the tech- nique (from paddle-and-anvil to wheel throwing). Moreover, the Spanish example will enable us to test the variability of ceramics made over the course of two days by the same potter.

Intra-group variability has been calculated on the basis of Coefficient of Variation (CV). CV is defined as the sample standard deviation divided by the sample mean, multiplied by 100 and expressed as a percentage. It may be considered as the standard statistic in studies of variation and therefore as an excellent measure of standardiza- tion (Eerkens and Bettinger 2001).


The results of these analyses enable us to assess how rate of production affects the metric variabil- ity for paddled and wheel-thrown pots.

Inter-Group Variabili~ in Low-Rate Production Contexts

Inter-group variability exists in low-rate production contexts; ANOVA shows significant differences @ < .001) between the series of vessels produced by the different Indian potters from Andhra Pradesh (Figures 3,4, and 5).Statistical tests (LSD [Scheffk] and Games-Howell) confirm these differences (p < .05).These differences hold for height, aperture, and maximum diameter."hey do not hold for the indexes of regularity of wall and of lip (width and thickness).

Pedda bana are produced less frequently than ralla catti and kura cntti. However, they are not characterized by inter-individual differences stronger than the ones displayed by the two other classes of vessels.

Aperture is established during wheel throwing and is not modified later by paddling, as are vessel height and maximum diameter. However, contrary

Kura catti






Figure 3. Inter-group differences among Andhra Pradesh potters (A-G) as shown by means and standard deviations of measurements taken on kura catli.

to what we expected, wheel throwing does not pro- duce less variability than the paddle-and-anvil tech- nique.

Inter-Group Variability in Higlz-Rate Production Contexts

North Indian vessels display inter-group variabil- ity as attested on the basis of ANOVA that shows significant differences (p < .001) between ghariya


Ralla catti

Pedda Bana 46 1




Figure 4. Inter-group differences among Andhra Pradesh potters (A-G) as shown by means and standard deviations of measurements taken on ralla catti.

produced by Delhi potters (Figure 6). Statistical tests (LSD [Scheffk] and Games-Howell) confirm these differences (p < .05). These differences hold for height, aperture, and maximum diameter.j

Intra-Group Variability in Low-Rate Production Contexts

The Andhra Pradesh ceramics yield the following





Figure 5. Inter-group differences among Andhra Pradesh potters (A-G) as shown by means and standard deviations of measurements taken on pedda bana.

results: variability in CVs for height, maximum diameter, and aperture is highest for pedda bana. For pedda bana, CV values range from 1.43 to

7.81 percent. For ralln catti, CV values range mainly from 2.01 to 4.70 percent, except for one potter whose CVvalues range from 6.61 to 7.15 per- cent. This potter (F) is the oldest (65 years old). For kura catti, CV values range mainly from 2.50 to

AMERICAN ANTIQUITY [Vol. 68,No. 4, 20031

Kura Catti

Ralla Catti

8. , 7-

potterl potter2 potter3 potter4 potter5 potter6

Pedda Bana
28.228 2potter1 potter2 potter3 potter4 potter5 potter6

Potters of New Delhi

Figure 6. Inter-group differences among New Delhi pot- ters (1-6) as shown by means and standard deviations of measurements taken on ghariya.

5.54 percent, except for one potter whose CV val- ues range from 5.98 to 9.71 percent (Figure 7). This potter (B), who is 35 years old, is acknowl- edged to not be as good a potter as his brother-in- law (A).

Whatever the absolute dimensions, whether height, maximum diameter, or aperture, standard deviations (SD) and CVvary within the same range. Since aperture is obtained by throwing, maximum diameter by paddling, and height by a combination of throwing and paddling, it follows that variation



Figure 7. Coefficients of variations (CV) of measurements taken on series of kura catti, ralla catti andpedda bana pro- duced by Andhra Pradesh potters (A-G).


Table 1. Measurements Taken on the Series of Ralla Catti Produced by Six Potters from Andhra Pradesh.

Max.   Wall Wall Wall Lip Lip
Potters Height Diameter Aperture Thick1 Thick2 Thick3 Width Thickness
Potter A (n= 31)                
Mean (cm) 13.13 20.8 17.42 .4 1 .46 .39 1.32  
SD .62 .68 .64 .07 .06 .09 .19  
CV (%) 4.7 3.29 3.68 16.22 13.54 22.76 14.57  
Potter B (n= 30)                
Mean (cm) 13.64 20.39 14.82 .25 .34 .52 1.83 .65
SD .57 .54 .58 .06 .08 .ll .ll .17
CV (%) 4.19 2.62 3.91 23.93 24.41 21.75 6.26 25.65
Potter D(n = 30)                
Mean (cm) 14.78 22.38 18.05 .40 .49 .54 1.76 1.4
SD .55 .56 .36 .06 .07 .07 .ll .15
CV (%) 3.75 2.5 1 2.01 14.9 13.58 13.48 6.44 10.5
Potter E (n= 30)                
Mean (cm) 14.47 22.79 17.02 .26 .31 .44 1.73 .53
SD .6 .56 .61 .05 .04 .06 .09 .07
CV (%) 4.16 2.45 3.58 21.01 13.66 14.08 5.47 12.39
Potter F (n= 18)                
Mean (cm) 13.84 22.23 17.06 .34 .38 .56 1.72 .94
SD .91 1.5 1.22 .05 .06 .07 .09 .14
CV (%) 6.61 6.73 7.15 13.98 16.95 12.38 5.38 14.68
Potter G(n = 27)                
Mean (cm) 14.68 22.11 17.25 .4 1 .39 .47 1.7 .74
SD .64 .84 .67 .08 .06 .07 .09 .13
CV (9%) 4.37 3.79 3.87 18.55 15.46 14.19 5.16 17.21

in dimension does not seem to depend on the fash- these pooled CV values higher than those for each ioning technique. potter. Amplitude of CVvaries, for some potters, from

Intra-Group Variability in High-Rate one category of vessel to the next. The same pot- Production Contexts

ter can present low CV values on pedda bana, and higher CV values on kura catti, or low CV values In Uttam Nagar (New Delhi), variability within on kura catti and higher CV values on ralla catti. potters' production is less than in Andhra Pradesh.

For the three types of vessels, thickness of walls Except for one variable (height of the jars made by shows very high intra-group variations, reflected in potter 5), all the CV values are less than 3 percent CVvalues that can reach 25 percent (Tables 1,2,3). (Table 5).Maximum diameter presents the lowest

CV values for lip thickness and width range values, ranging from .87 to 1.40 percent. CV val- mostly from 4 to 8 percent. However, some vessel ues of aperture range from 1.19 to 2.83 percent. CV groups show high intra-individual variation (Tables values of height range from 1.40 to 3.19 percent. 1,2, 3). When pooling the potters, height, maximum

When considering the standard deviations and diameter and aperture present CVs that range from CVvalues of each type of vessel, all potters pooled, 1 to 5 percent (Table6). Once again the cumula- it appears that, for height, maximum diameter and tive effect results in CV values higher than those aperture, CVvalues range from 5 to 9 percent (Table for each individual potter. 4). There is a cumulative effect, also called "cumu- In Spain, whatever the series considered (A, B lative blurring" (Blackman et al. 1993) that makes or C), the dimensions that vary least are height,

AMERICAN ANTIQUITY [Vol. 68, No. 4, 20031

Table 2. Measurements Taken on the Series of Kzim Catti Produced by Seven Potters from Andhra Pradesh.

Max. Wall Wall Wall Lip Lip

Potters Height Diameter Aperture Thick1 Thick2 Thick3 Width Thickness Potter A (11 = 30) Mean (cm) 14.11 19.82 13.7 .3 .53 .4 2.73 1 .O SD .55 .78 .57 .08 .10 .08 .24 .14 CV (70) 3.87 3.94 4.19 25.95 18.06 20.85 8.66 13.81

Potter B (n = 30) Mean (cm) 14.66 20.01 14.04 .37 .38 .44 2.98 1.18 SD .88 1.94 .87 .07 .07 .06 .24 .I10 CV (%) 5.98 9.71 6.21 18.0 18.23 13.9 7.88 9.39

Potter C (n = 11) Mean (cm) 15.48 21.17 14.82 .30 .55 .45 2.77 1.1 SD .45 .86 .55 .07 .09 .06 .18 .10 CV (7G) 2.88 4.06 3.68 22.59 16.49 13.1 6.41 9.5

Potter D (n = 3 1) Mean (cm) 15.34 20.49 14.3 .30 .52 .52 2.58 1.07 SD .47 .51 .43 .04 .07 .07 .ll .12 CV (70) 3.08 2.5 3.0 14.66 14.12 14.12 4.13 11.44

Potter E (n = 30) Mean (cm) 15.03 20.96 15.23 .21 .36 .39 2.74 1.01 SD .56 .68 .59 .06 .05 .04 .ll .08 CV (%) 3.71 3.26 3.9 29.48 12.84 10.15 4.14 7.95

Potter F (n = 24) Mean (cm) 14.37 21.09 15.28 .27 .34 .37 2.25 1.12 SD .63 .96 .85 .06 .08 .06 .I3 .09 CV (%) 4.38 4.55 5.54 23.24 22.62 15.39 5.85 7.88

Potter G (n = 30) Mean (cm) 15.27 20.85 15.22 .29 .48 .49 2.14 .89 SD .65 .70 .58 .05 .08 .06 .15 .07 CV (70) 4.26 3.35 3.82 18.05 16.17 11.74 7.01 7.44


maximum diameter, and aperture (rim and neck diameter). Their CV values range from 1.4 to We can now examine whether rate of production

2.9 percent (Table 7). Such a variation in the pre- affects motor habits and consequently degree of cision of the dimensions has been observed for the standardization. production of six other workshops located in dis- Our results suggest that the series of vessels tant villages. The other dimensions present a greater considered here present different degrees of stan- degree of variability (higher CV). dardization resulting from differing rates of pro-

The standard deviation of series A is smaller than duction. High-rate ceramic production is less that of series B, which means that the potter made variable than low-rate production (Tables 4,6, and vessels with a higher degree of variability on day 7). However, even with high rates of production, B than on day A. When statistically comparing the work units may be distinguished through metric means (Aspin Welsh test), it appears that means dis- variables as exemplified by the inter-group analy- tinguish the two groups for the following parame- sis of New Delhi potters (Figure 6). ters: capacity, weight, height of maximum diameter, Variability of ceramic assemblages is best and height of the decoration. They are the para- expressed by CVs that are placed here along a con- meters with the higher CV. tinuum of variation from .87 to 25percent. The


Table 3. Measurements Taken on Series of Pedda Bana Produced by Six Potters from Andhra Pradesh.

Max Wall Wall Wall Lip Lip

Potters Height Diameter Aperture Thick1 Thick2 Thick3 Width Thickness Potter A (n = 12) Mean (cm) 39.09 35.68 16.94 .40 .48 .42 4.12 1.69 SD 2.45 2.34 .77 .08 .08 .09 .25 .14 CV (70) 6.27 6.56 4.55 20.06 15.7 21.77 6.11 8.15

Potter B (n = 16) Mean (cm) 40.08 37.48 16.56 .44 .69 .52 4.35 1.78 SD 1.90 1.17 .70 .07 .ll .08 .21 .21 CV (5%) 4.75 3.11 4.20 15.06 16.13 15.72 4.85 12.02

Potter D (n = 16) Mean (cm) 38.58 36.74 17.91 .55 .80 .69 1.83 1.14 SD .84 1.24 .7 1 .08 .08 .08 .I0 .10 CV (a) 2.17 3.38 3.96 14.19 10.21 11.11 5.54 8.43

Potter E (n = 15) Mean (cm) 42.58 40.11 19.15 .43 .55 .55 3.58 1.46 SD 2.25 1.55 1.35 .09 .07 .05 .16 .10 CV (a) 5.29 3.86 7.04 21.18 11.9 9.94 4.50 6.75

Potter F (n = 13) Mean (cm) 37.18 35.61 15.35 .48 .5 1 .46 4.44 1.78 SD 2.9 2.16 .77 .09 .07 .I 1 .33 .13 CV (9%) 7.81 6.05 5.04 17.96 13.33 23.05 7.33 7.18

Potter G (n = 13) Mean (cm) 38.21 35.93 15.8 .47 .73 .53 3.44 1.44 SD .55 .77 .74 .06 .08 .07 .17 .09 CV (5%) 1.43 2.14 4.7 13.44 10.28 12.48 4.97 6.05

minimum CVis below 1.7 percent, which is the CV been observed (e.g., Benco 1988) although some value derived from the Weber fraction6 (Eerkens ethnographic cases tend to show the opposite and Bettinger 2001), although no automation or (Arnold and Nieves 1992). It can be considered independent standard was used in this case. Here, robust since we could isolate the technical factor we assume that it is possible to obtain such high by analyzing pots made according to different degrees of standardizationgiven a high degree of

Table 4, Measurements Taken on the Series of Vessels

motor expertise acquired through an optimal per-

Produced by the Potters from Andhra Pradesh. ception-action ongoing cycle (Bril et al. 2000). Height, maximum diameter, and aperture are Type of vessel Height Max. Diameter Aperture the variables most likely to vary according to motor R,II, catti (n = 166) habits. Lips are stylistic features that allow potters Mean (cm) 14.1 SD .88

to express themselves as they please. In Spain, the

CV (%) 6.27

variation in the base diameter could be related to the absence of visual reference when manufactur-

Kura catti (n = 174) ing the base in the course of the hollowing opera- Mean (cm) 14.81 tion. Variation in height of decoration and turning SD .78 CV (%) 5.23

zone corresponds to the fact that they are secondary attributes that do not contribute to the general shape

Pedda bana (n = 84)

of the pot.

Mean (cm) 39.37 The forming technique does not affect variabil- SD 2.58

ity of ceramic assemblages. This result has already CV (5%) 6.56

AMERICAN ANTIQUITY [Vol. 68, No. 4,20031

Table 5. Measurements Taken on the Series of Vessels 
Produced by the Six Potters from Uttam Nagar (New 


Potters Height Diameter Aperture
Potter 1 (n= 30)      
Mean (cm) 30.36 29.49 11.9
SD .58 .36 .22
CV (7~) 1.9 1.22 1.87
Potter 2 (n = 30)
Mean (cm) 30.5 1 29.19 12.36
SD .48 .32 .I5
CV (%) 1.56 1.08 1.19
Potter 3 (n = 30)
Mean (cm) 30.3 28.82 1 1.64
SD .61 .40 .26
CV (%) 2.01 1.40 2.25
Potter 4 (n = 30)
Mean (cm) 28.72 29.3 13.09
SD .69 .36 .31
CV (%) 2.41 1.24 2.34
Potter 5 (n= 30)
Mean (cm) 28.19 28.74 11.52
SD .90 .37 .33
CV (70) 3.19 1.30 2.83
Potter 6 (n = 30)
Mean (cm) 30.37 28.69 12.14
SD .42 .25 .29
CV (%) 1.40 .87 2.41

techniques (wheel throwing vs. paddle-and-anvil).

In Andra Pradesh, there is a tendency for the biggest pots to be less standardized than smaller pots. This tendency has been observed elsewhere (see in particular Benco 1988). It may be that errors in estimating object size increase linearly with the magnitude or size of the intended end product (e.g., Coren et al. 1994 cited in Eerkens and Bettinger 2001 :494). Another explanation could be that the biggest pots are produced in smaller quantities and, therefore, that motor habits are less practiced than those for smaller pots. Therefore, metric compar- isons should take size classes into account (as underlined also by Stark 1995).

Skill may affect the range of variability. A skill may be defined as the capacity to reach a goal through use of the resources in one's environment (Bril et al. 2000). This capacity may vary from one person to the other, as shown by the Andhra Pradesh

Table 6. Measurements Taken on the Series of Vessels 
Produced by the Potters from Uttam Nagar (New Delhi). 

Ghariya (n= 180) Height Max. Diameter Aperture
Mean (cm) 29.74 29.04 12.11
SD 1.12 .47 .59
CV (%) 3.77 1.61 4.85

potters, among whom some, of the same age, have the same amount of experience. Metric data from San Nicolas in the northern Philippines (Longacre 1999) support this result. Pots produced by the less- skilled potter show greater metrical variation.

The different results obtained for Indian and Spanish pottery can be compared with those from the Philippines. The hypothesis is that if motor habits determine degree of standardization, then we should have, for the three different cultures (Indian, Spanish, Filipino), the same range of metric vari- ability for series of vessels produced within the same range of production (low-rate vs. high-rate production). In other words, intercultural compar- ison should enable us to assess how ceramic vari- ability may be affected by different emic conceptions of standardization. In Andhra Pradesh, the size-class definition is a function of the size of the family that will use the vessels. In the Philip- pines, the size classes are based on vessel volume, reckoned through the chupa (ca. 100 ml)(Kvamme et al. 1996). In Spain, standardization indexes degree of skill and the potter's intention is to make very similar pots, in contrast to Uttam Nagar's pot- ters, whose vessel measurements respond to mar- ket-oriented norms. In the Indian and Spanish contexts, all potters pooled, high-rate production yielded ceramic assemblages whose CVvalues are less than 5 percent, whereas low-rate production yielded ceramic assemblages whose CV values range from 5 to 9 percent. These results are slightly different from the ones obtained in the Philippines (Kvamme et al. 1996). In the Philippines, all pot- ters pooled, the high-rate production (Paradijon, full-time potter specialists) presents CVvalues that range from 4.3 to 6.92 percent, whereas low-rate production (Dalupa, part-time potter specialists) presents CVs ranging from 4.06 to 5.56percent. The least-specialized potters (Dangtalan, household use, and restricted exchange) produce ceramics whose CVvalues range from 6.27 to 8.34 percent. Paradijon and Dalupa ceramics differentiate on aperture only (B-F test,p = ,0482) whereas Dalupa


Table 7. Measurements Taken on the Series of Pitchers 
Produced by One Potter in the Course of Two Days. 

Measurements Series Mean (mm) SD CV (%) Maximum diameter A 121.1 1.7 1.4 B 120.2 2.4 2.0 C 121.0 2.1 1.7 Aperture A 98.9 2.1 2.2 B 100.1 2.4 2.4 C 100.1 2.6 2.5 Neck diameter A 82.7 1.9 2.3 B 80.9 2.2 2.6 C 82.1 2.2 2.7 Base diameter A 49.6 2.6 5.7 B 50.4 2.2 4.4 C 50.0 2.3 4.6 Total height A 78.8 2.1 2.6 B 80.4 2.2 2.6 C79.3 2.4 2.9 Maximum diam. A 40.3 1.8 4.4 height B 40.6 2.1 5.1 C 41.1 1.9 3.7 Lip height A 13.9 .9 6.4 B 14.4 .6 4.3 C 14.4 .7 4.9 Lip thickness A 5.0 .1 2.0 B 5.9 .3 6.0 C 5.0 .2 4.8 Handle width A 17.5 .7 4.0 B 17.2 .5 3.0 C 17.4 .8 2.8 Handle thickness A 10.5 .5 4.7 B 9.8 .5 5.1 C 10.2 .6 6.0 Height of the decor A 46.9 2.3 4.9 B45.9 3.1 6.8 C 46.1 2.9 6.1 Height of the A 56.4 5.2 9.2 non-turned zone B 51.8 5.8 11.2 C 55.2 6.0 10.0 Capacity A 489.6 25.1 5.1 B 490.2 18.5 6.2 C 492.1 28.3 15.7Weight A 353.2 14.5 4.1 B 348.5 17.4 5.0 C 352.5 16.8 4.7 Note: Series A (n = 50) and B (n = 50) were made over two days. Series C pools series A and B. Data from Arcellin- Pradelle and Laubenheimer (1985).

and Dangtalan ceramics differentiate on aperture and height (B-Ftest,p =.008,p=,0003) (Kvamme et al. 1996). The three communities employ a com- bination of coil-and-scrape and paddle-and-anvil techniques to fashion their pots.

These observations suggest, first, that small- scale production does develop motor habits that enable potters to produce standardized series of vessels comparable to those made by high-rate pro- ducers. This point is supported by the San Nicolas data published by Longacre (1999). Series of water jars (62,36,25, and 20 pots made respectively by four potters) vary between about 3 and 4 percent metrically. Their annual production is comparable to that achieved by the Andhra Pradesh potters (Longacre indicates that they make 100 pots per week when it is not raining). Small-scale produc- tion refers here to annual production of a few thou- sands pots. It has to be distinguished from very small-scale production whose metric variability is much higher. The Kalinga pottery varies about 12 percent (Longacre 1999). Such variation is com- parable to the variation observed in Papua andAma- zonia for very low production (Stark 1995).

Second, emic conceptions of class sizes may induce a certain variability, which is why CV val- ues do not range the same way from one culture to the other, despite similar rates of production (as observed also by Longacre 1999).

Third,in a cumulative-effect situation, no small- scale ceramic assemblage presents at least two vari- ables with CV values lower than 3 percent.

To summarize, the ethnographic data presented here suggest that ceramic assemblages presenting CV values from 3 to 6 percent are quite ambigu- ous. They may result from large-scale as well as small-scale production. Ceramic assemblages pre- senting CV values from 6 to 9 percent may belong to small-scale as well as very small-scale produc- tion. Ceramic assemblages presenting at least two variables with CV values below 3 percent point to high-scale production (Figure 8).

These ceramic assemblages are made up of fewer than 10 production events (i.e., let us recall that a production event corresponds to a potter's production). In the future, more assemblages rep- resenting more numerous production events should be analyzed. Indeed, we have seen that the more we pool potters together, the greater the cumula- tive effect that plays on the CVvalues. This cumu- lative effect should be amplified if assemblages are made up of production events achieved over a few centuries, or else multiple production events.

Finally, let us recall that large-scale production is achieved by full-time specialists, whereas low- scale production may be achieved either by full- time or part-time specialists, or even by the least-specialized potters in a community.

Large-scale production (14,000 pots a year per potter)

< 3%




Large-scale production 
or small-scale production 
(between 14,000 and 4,000 pots 
a year per potter) 

3% -6%


[Vol. 68, No. 4, 20031

Small-scale production or 
very small-scale production 
(6,000pots a year or less 
per potter) 

a> 6%

AndIua Pradesh

Fiyre 8. Quantifying rate of production by reference to ceramic assemblages made up of less than 10 production events.

Application To Archaeological Data

Let us now apply our results to archaeological data. We consider the "open-simple-rim" fine ware bowls that had become one of the most characteristic ceramic forms by 2200 B.C. at Tell Leilan and numerous other sites in Northern Mesopotamia. As outlined by Blackman et al. (1993), they offer the rare opportunity to examine the notion of stan- dardization since large waster stacks have been recovered in the eastern part of the Leilan lower town (Period IIb). A portion of the stacks, con- taining 27 whole bowls and bowl fragments, was brought to United States and examined by Black- man et al. (1993). According to their results, stan- dardized procedures were followed at every step of the bowl-production process. Chemical analysis of the bowls in the waster stack shows a remarkably high degree of homogeneity in the chemical com- position of the constituent clays. Concerning ves- sel dimensions, five measurements were made: rim diameter, wall thickness 1 cm below the rim, ves- sel height, base diameter, and maximum basal thickness. According to Blackman et al. (1993), these bowls show a high degree of standardization since CVvalues are less than 10 percent. From this, they conclude that "metric data from the Leilan wasters show a level of homogeneity consistent with expectations for specialized mass production of utilitarian goods" (Blackman et al. 1993:73).

Tell Leilan bowls are supposed to be a single production event. If we compare Tell Leilan bowls with Indian, Spanish, and Filipino vessels in terms of standard deviation and CV, it appears that we are dealing with a weakly standardized produc- tion system: CV values of rim diameters are

9.19 percent and CVvalues of heights are 4.4 per- cent. These CV values, which are not affected by any cumulative effect, are high compared to our ethnographic reference points. They do not corre- spond to so-called large-scale production. They may reflect small-scale production, which means that, even if they were made by specialists (as attested by the skills required for their fashioning; see Roux and Corbetta 1989), the rate of produc- tion may not have been so high (Figure 8).

The low degree of standardization that appears in this production is quite consistent with the vari- ability measured on a second series of open-sim- ple-rim bowls recovered from domestic refuse contexts (Leilan, Period IIb). This second series could have been produced over 200 years. CVval- ues reach 18 percent. Blackman et al. (1993) inter- pret this variability in terms of noncentralized production; i.e., multiple workshops during period IIb. According to our ethnographic data, the cumu- lative effect of intra- and inter-group variability should not be underestimated (as underlined also by Blackman et al. 1993). In the case of the sec- ond series of bowls from Tell Leilan, the 18


~ercentCV values could be the result of a curnula- tive effect over 200 years and correspond to ves- sels whose degree of variability, for a single production event, could not have exceeded 9 percent. The variability of the two series of bowls may therefore be highly comparable (as suggested also by Blackman et al. 1993:74) and interpreted in terms of similar rates of production.


In this paper, we have examined the standardiza- tion hypothesis through the metric variability pre- sented by series of vessels produced in different economic contexts. By comparing different eco- nomic situations belonging to the same culture (Indian), we have been able to isolate the parame- ter "intensity of production" in order to test the hypothesis according to which rate of production plays on motor habits that, in turn, affect the degree of standardization. Our results show that rate of pro- duction does affect the degree of standardization. However, when comparing ceramic assemblages from different cultures (Filipino), it appears that emic conceptions of standardization also play on metric variability. Only in a high-rate production situation do we have motor habits that transcend emic conceptions of standardization. It enables us to quantify, even if hypothetically, the relative range of variability expected on vessels made within dif- ferent contexts of production. When applied to archaeological data, these results show that some ceramic production considered up to now as highly standardized may correspond, in fact, to relatively small-scale production.

By way of conclusion, let us emphasize that archaeological data need to be interpreted by ref- erence to quantified ethnographic data. Quantifi- cation has to be cross-cultural so that it can be applied in any cultural context. In order to elabo- rate cross-cultural referents or regularities, techni- cal facts have to be studied in terms of resources and constraints (Roux 1999).The constraints con- sidered here are the motor habits underlying the production of standardized series of vessels. The cultural aspect is taken into account through the emic conceptions of standardization. One of the aims of ethnoarchaeological study is also to assess the context into which constraints transcend emic conceptions. Ultimately, such an approach enables us to test our interpretations against empirical data.

Ackno~vledgments.This research was supported by a grant from the French Institute of Pondicherry (Ministry of Foreign Affairs). A first version of the paper was presented at the 1994 World Archaeological Congress in New Delhi, India. I thank J. Pouchepadass for support of the project and

M.L.K. Murty for his precious collaboration in the field. Comments on earlier drafts of this paper by Patricia L. Crown and anonymous reviewers are especially acknowl- edged. I also kindly thank Tim Kohler for critique, sugges- tions, and editorial comments. I am grateful to Jehanne FCblot-Augustins and Carol Kramer for their help in improv- ing my English. Figures were drafted by Gerard Monthel (CNRS, UMR7055). The Spanish abstract was kindly trans- lated by Jean-Fran~ois Bouchard.

References Cited

Arcelin-Pradelle, C., and F. Laubenheimer

1982 Une Mission d'Arch6o-Ethnologie: les Ateliers de Potiers Traditionnels dans les Provinces de Valencia et de Castellin (Espagne). Techniq~ieset Technologies. Sozirces Docutnentaires 5:27-52. Institut de Recherches MCditer- ranCennes, UniversitC de Provence.

1985 La Notion de Serie de Production en Ceramique TournCe. Techniques et Technologies. Sources Docurnen- taires 7:129-139. Institut de Recherches MCditer- raneennes, UniversitC de Provence.

Arnold. D.. and A. L. Nieves 1992 Factors Affecting Ceramic Standardization. In

Ceramic Production and Distribution: an Integrated Approach, edited by G.J. Bey 111 and C.A. Pool, pp. 113-214. Westview Press, Boulder. Colorado.

Benco, N.

1988 Morphological Standardization: An Approach to the Study of Craft Specialization. In A Pot for All Reasons: Ceramic Ecology Revisited. edited by C. Kolb and L. Lackey. pp. 57-72. Temple University, Philadelphia.

Blackman. M. J., G. J. Stein, and P. B. Vandiver

1993 The Standardization Hypothesis and Ceramic Mass Production: Technological, Compositional and Metric Indexes of Craft Specialization at Tell Leilan, Syria.Atnerican Antiqziiy 58:60-8.

Bril. B., V. Roux, and G. Dietrich

2000 Skills Involved in the Knapping of Chalcedony Beads: Motor and Cognitive Characteristics of a Complex Situ- ated Action. In Cornaline de I'lnde: Des prariques tech- niqries de Carnbay ~LLXtechno-systemes de I 'lndus,edited by V. Roux, pp. 207-332. Editions de la MSH, Paris (Eng- lish CD-Rom enclosed).

Coren, S., L. M. Ward, and J. T. Enns 1994 Sensarion and Perception. 4th ed. Harcourt Brace, Fort Worth, Texas. Costin, C. L.

1991 Craft Specialization: Issues in Defining, Document- ing and Explaining the Organization of Production. In Ad~ances in Archaeological Method and Theoq, Vol. 3, edited by M. B. Schiffer. pp. 1-56. University of Arizona Press, Tucson.

2000 The Use of Ethnoarchaeology for the Archaeological Study of Ceramic Production. Jo~ir-naluf Archaeological Method and Theory 7:377403.

Costin. C. L.. and M. B. Hagstrum

1995 Standardization. Labor Investment, Skill, and the Organization of Ceramic Production in Late Prehispanic Highland Peru. American Antiq~tit). 60:619-639.

AMERICAN ANTIQUITY [Vol. 68, No. 4,20031

Eerkens, J. W., and R. L. Bettinger

2001 Techniques for Assessing Standardization in Artifact Assemblages: Can We Scale Material Variability? American Antiquity 66:493-504.

Feinman, G. M., S. Upham, and K. G. Lightfoot

1981 The Production Step Measure: An Ordinal Index of Labor Input in Ceramic Manufacture. American Antiquity 46:871-884.

Gallay, A. 1986 Lhrcheologie Denzain. Belfond, Paris. Kvamme. K. L., M. T. Stark, and W. A. Longacre 1996 Alternative Procedures for Assessing Standardization

in CeramicAssemblages.A1?~ei-icanAntiqui~61:116-126.

3. The Least Significant Difference (LSD) is the mini-

London, G. A. 1991 Standardization and Variation in the Work of Craft Specialists. In Ceramic Ethnoarchaeology, edited by W.A. Longacre. pp. 182-204. University of Arizona Press. Tuc- son.

Longacre, W. A. 1991 Sources of Ceramicvariability among the Kalinga of Northern Luzon. In Ceramic Ethnoarchaeology, edited by

W.A. Longacre, pp. 95-1 1. University of Arizona Press, Tucson.

1999 Standardization and Specialization: What's the Link? In Pottery and People, edited by J. Skibo and G. Feinman, pp. 23 1-267. University of Utah Press, Salt Lake City.

Longacre, W. A,,K. L. Kvamme. and M. Kobayashi 1988 Southwestern Pottery Standardization: An Ethnoar- chaeological View from the Philippines. Kiva 53: 101-1 12. Rice, P. 1989 Ceramic Diversity, Production and Use. In Quantifiing Diversity in Archaeology, edited by R. D. Leonard and

G. T. Jones, pp. 109-117. Cambridge University Press, Cambridge.

1991 Specialization, Standardization and Diversity: a Ret- rospective. In The Ceramic Legacy of Anna 0.Shepard, edited by R. L. Bishop andF. W. Lange, pp. 257-279. Uni- versity Press of Colorado. Boulder, Colorado.

Roux, V., and D. Corbetta 1989 The Potter's Wheel: Craft Specialization and Techni- cal competence. Oxford and IBH Publishing, New Delhi. Roux, V. 1999 Ethnoarcheology and the Generation of Referential Models: The Case of Harappan Carnelian Beads. In Ethno-Analogy and the Reconstruction ofPrehistoric Artefact Use and Production, edited by L. Owen and M. Porr, pp. 153-17. MoVince Verlag, Tiibingen.

Saraswati, B., and N. K. Behura 1964 Potren Techniques in Peasant India. Anthropological Survey of India, Vol. 13, Calcutta. Sinopoli, C. 1988 The Organization of Craft Production at Vijayana- gara, South India. American A~~thropologist

90:580-597. Stark, B. L. 1995 Problems in the Analysis of Standardization and Spe-

cialization in Pottery. In Ceramic Production in the Amer-

ican South~vest. edited by B. Mills and P. Crown, pp.

231-267. University of Arizona Press, Tucson.

  1. The ethnoarchaeological survey was carried out in Andhra Pradesh, in February 1991, in collaboration with Prof. M.L.K. Murty (University of Hyderabad).


  2. Index of regularity of walls = (ThicklIL-ThickliR)2 + (Thick2L-Thick2/R)2 + (Thick3L-Thick3/R)2. (L = left, R = right). Likewise for thickness and width of the lip.


mum difference between the two groups that can be consid- ered statistically significant.

  1. For kura carri (Figure 3), height differentiates between potter A and potters C, D, E, G, potter F and potter C. D. E. G, and potter B and potters C, D, G. Maximum diameter dif- ferentiates between potter A and potters C, D, E, F, G. and potters D and E. Aperture differentiates between potter A and potters C, D. E, F, G, potters B and C, E, F, G, and potters D and E, F, G (Games-Howell test). For ralla carti (Figure 4), height differentiates potter A and potters B, D, E, G, potter B and potters D, E, G, and potter F and potters D and G. Maximum diameter differentiates between potter B and pot- ters D, E, F, G, potter A and potters D, E, F, G, and potters G and E. Aperture differentiates between potter B and potters A, D, E, F, G, and potter D and potters A, E. F, G (Games- Howell test). For pedda bana (Figure 5), height differentiates between potter F and potters B, E, potter G and potters B, E, and potter E and potters A, B. D. Maximum diameter differ- entiates between potter E and potter A, B, D, F, G, and pot- ters G and B. Aperture differentiates between potter F and potters A, B, D, E, potter G and potters A, D, E, potter B and potters D, E, potter A and potters D, E, and potters D and E (Games-Howell test).


  2. Height differentiates between potters 5 and 1, 2, 4. 6. potters 3 and 1, 2, 4, 6, potters 1 and 2, 4, 6, potters 6 and 4, and potters 2 and 4. Aperture differentiates each potter from each other. Maximum diameter differentiates between two groups of potters: potters 3, 5, 6 and potters 1, 2, 4 (Games- Howell test).


  3. "Unlike mechanical scales that determine weight within an invariant unit of error, human appreciation of heav- iness is scaled relative to object weight. This value (2 percent) has come to be called the Weber's fraction for heaviness" (Eerkens and Bettinger 2001:495).


Received March 25, 2002; Revised October 31, 2002; Accepted April 17, 2003.

  • Recommend Us