Wrought Metal-Working Prior to Middle Shang …earlychina/docs/earlychinajournal/ec6_1_barnard.pdf4...

27
4 WROUGHT METAL-WORKING PRIOR TO MIDDLE SHANG (?)—A PROBLEM IN ARCHAEOLOGICAL AND ART-HISTORICAL RESEARCH APPROACHES NOEL BARNARD Department of Far Eastern History Australian National University GPO Box 4, Canberra ACT 2600, Australia During the course of the Conference on "The Origins of Chinese Civilization" held at Berkeley in June, 1978, two papers made reference to an assumed use of wrought metal-working techniques in Middle Shang, Early Shang (or Hsia?), and Ch'i-chia 齊家 Cultural contexts. 1 The main proponent of this theory was Mr. Robert W. Bagley. As a consider- able volume of research which has important bearing on the origin and the development of metallurgy in China has appeared in recent years from the hands of such authorities as Orvar Karlbeck, Li Chi 李濟 Shih Chang-ju 石璋如 , Wilma Fairbank, John Gettens, Wan Chia-pao 萬家保 ,as well as the several studies conducted by the present author — in partic- ular (with Satō Tamotsu 左藤保 the recently published Metallurgical Remains of Ancient China (1975), it would seem to be essential that the al- leged evidence of smithy work at so remote a period be reviewed critically and in some detail. Up to now there has been little need to embark upon a detailed appraisal of the subject for the very good reason that controlled archaeological explorations on the Chinese mainland have presented no signs at all of the applications of metal-working throughout many thousands of excavated metal artifacts. 2 The purpose of this short study is to clarify a situation that has become rather confused because of several misunderstandings of the significance of certain features in ceramic and in bronze-cast ves- sels; these have been incorrectly interpreted as evidence of copying from metal prototypes. Bagley is not alone in falling into this trap, nor was he the first to do so. Several scholars have earlier commented to the effect that certain types of Chinese pottery vessels appeared (to them) to be based upon metal prototypes, e.g., M. Bylin-Althin (BMFEA 15 [1946], pp. 426-427) in regard to Ch'i-chia-p'ing 齊家坪 ware: In their design all the vessels just referred to appear to be traceable to metal prototypes, although otherwise no traces of any metal cul- ture have been found in Ch'i-chia-p'ing.... The fact that the handle is attached to a portion projecting from the over-vaulted rim, this part being decorated with two knobs resem- bling rivet-heads, lends force to the general characterization of the vessel as a direct imi- tation of metal ware. A sketch of the vessel just referred to appears in Figure 1. 3 An Chih-min 安志敏 , in a study on the material culture of the Chinese Neolithic (WW 1956.8:46), refers to the li-ho 鬲盉 style pottery vessels of Lung-shan date. These, he believes, were patterned on metal prototypes and accordingly indi- cate that the Lung-shan Culture had entered the Metal Age although direct evidence in the form of metal artifacts had not yet been unearthed. He refers to the two protruberances on the handles "emulating rivets" (模仿鉚釘 ), but otherwise (like Bylin- Althin) does not specifically imply that the assumed prototypes would have been fashioned by smithy techniques. 4 Probably Sekino Takeshi 關野 , ("Ryusanbunka no kaimei" 龍山文化の解明 , Sundai Shigaku 駿台史学 11 [1961]:1-22), was the first to opine specifically that not only were the bronze versions of the li-ho manufactured by smithy techniques but also that evidence of such smithy manufactured metal prototypes could be observed in Lung-shan ceramic wares. His discussion on the latter point is based on the Andersson Ch'i-chia style ho- kettle (Figure 1) and details of his argument have in effect (and, no doubt, fortuitously), been dupli- cated by Bagley in his recent survey. It is not sur- prising that these consultations of the article by Bylin-Althin and the repeated studies of the Andersson ho-kettle should each result in much the same con- clusions. In Metallurgical Remains (p. 8, fn. 9) I refuted Sekino's conclusions that the bronze li-ho kettles were manufactured largely by smithy tech- niques. 5 One of the three vessels upon which his study was based is in the Avery Brundage Collection and I had personally examined it (B60.B1078; and also the second one, B60.B53). Max Loehr in his notes on the Brundage bronze li-ho (B60.B53) in Ritual Vessels of Bronze Age China (1968), p. 30 remarks: "Still farther west in Kansu Province, pottery spouted vessels with domed top and tubular spout on baluster-shaped bodies with handles evoking metal prototypes, closely resembling the

Transcript of Wrought Metal-Working Prior to Middle Shang …earlychina/docs/earlychinajournal/ec6_1_barnard.pdf4...

4

WROUGHT METAL-WORKING PRIOR TO MIDDLE SHANG (?)—A PROBLEM IN ARCHAEOLOGICAL AND

ART-HISTORICAL RESEARCH APPROACHES

NOEL BARNARD Department of Far Eastern History Australian National University

GPO Box 4, Canberra ACT 2600, Australia

During the course of the Conference on "The Origins of Chinese Civilization" held at Berkeley in June, 1978, two papers made reference to an assumed use of wrought metal-working techniques in Middle Shang, Early Shang (or Hsia?), and Ch'i-chia 齊家 Cultural contexts.1 The main proponent of this theory was Mr. Robert W. Bagley. As a consider-able volume of research which has important bearing on the origin and the development of metallurgy in China has appeared in recent years from the hands of such authorities as Orvar Karlbeck, Li Chi 李濟Shih Chang-ju 石璋如, Wilma Fairbank, John Gettens, Wan Chia-pao 萬家保,as well as the several studies conducted by the present author — in partic-ular (with Satō Tamotsu 左藤保 the recently published Metallurgical Remains of Ancient China (1975), it would seem to be essential that the al-leged evidence of smithy work at so remote a period be reviewed critically and in some detail. Up to now there has been little need to embark upon a detailed appraisal of the subject for the very good reason that controlled archaeological explorations on the Chinese mainland have presented no signs at all of the applications of metal-working throughout many thousands of excavated metal artifacts.2

The purpose of this short study is to clarify a situation that has become rather confused because of several misunderstandings of the significance of certain features in ceramic and in bronze-cast ves-sels; these have been incorrectly interpreted as evidence of copying from metal prototypes. Bagley is not alone in falling into this trap, nor was he the first to do so.

Several scholars have earlier commented to the effect that certain types of Chinese pottery vessels appeared (to them) to be based upon metal prototypes, e.g., M. Bylin-Althin (BMFEA 15 [1946], pp. 426-427) in regard to Ch'i-chia-p'ing 齊家坪 ware:

In their design all the vessels just referred to appear to be traceable to metal prototypes, although otherwise no traces of any metal cul-ture have been found in Ch'i-chia-p'ing.... The fact that the handle is attached to a portion projecting from the over-vaulted rim,

this part being decorated with two knobs resem-bling rivet-heads, lends force to the general characterization of the vessel as a direct imi-tation of metal ware.

A sketch of the vessel just referred to appears in Figure 1.3 An Chih-min 安志敏, in a study on the material culture of the Chinese Neolithic (WW 1956.8:46), refers to the li-ho 鬲盉 style pottery vessels of Lung-shan date. These, he believes, were patterned on metal prototypes and accordingly indi-cate that the Lung-shan Culture had entered the Metal Age although direct evidence in the form of metal artifacts had not yet been unearthed. He refers to the two protruberances on the handles "emulating rivets" (模仿鉚釘), but otherwise (like Bylin-Althin) does not specifically imply that the assumed prototypes would have been fashioned by smithy tech niques.4 Probably Sekino Takeshi 關野雄, ("Ryusanbunka no kaimei" 龍山文化の解明, Sundai Shigaku 駿台史学 11 [1961]:1-22), was the first to opine specifically that not only were the bronze versions of the li-ho manufactured by smithy techniques but also that evidence of such smithy manufactured metal prototypes could be observed in Lung-shan ceramic wares. His discussion on the latter point is based on the Andersson Ch'i-chia style ho-kettle (Figure 1) and details of his argument have in effect (and, no doubt, fortuitously), been dupli-cated by Bagley in his recent survey. It is not sur-prising that these consultations of the article by Bylin-Althin and the repeated studies of the Anders son ho-kettle should each result in much the same con-clusions. In Metallurgical Remains (p. 8, fn. 9) I refuted Sekino's conclusions that the bronze li-ho kettles were manufactured largely by smithy tech-niques.5 One of the three vessels upon which his study was based is in the Avery Brundage Collection and I had personally examined it (B60.B1078; and also the second one, B60.B53).

Max Loehr in his notes on the Brundage bronze li-ho (B60.B53) in Ritual Vessels of Bronze Age China (1968), p. 30 remarks: "Still farther west in Kansu Province, pottery spouted vessels with domed top and tubular spout on baluster-shaped bodies with handles evoking metal prototypes, closely resembling the

5

Fig. 1. The pottery ho purchased by Andersson in Lan-chou 蘭州 (K5523), after BMFEA 15(1943), Pl. 38. It has been assumed that the flap-like element in the upper section of the handle, with the two small dome-shaped pro-truberances, imitate features extant in metal prototypes. This vessel, and others like it, are accordingly assumed to have been copied by ancient potters conversant with such metal vessels. These metal vessels, it has further been assumed, would have been made by smithy techniques; the two "rivets" are regarded as part proof. So, too, the extended flap, though this feature is not so pronounced in other pottery versions.

Fig. 2. Middle Shang type bronze li-ho kettlesof a design which is considered to be based upon earlier smithied versions. Features attending these hypothetical vessels (neces-sarily of Lung-shan and Early Shang date) are thought to be preserved in whole, or in part, in the relevant pottery forms in Figs. 1 and 3; and in the cast bronze examples above. 1: Li-ho recently acquired by the T'ien-chin

天津 Cultural Bureau (after WW 1964.9:33,Pl. ?, 1); rubbing of decor band below. 2: The Berlin State Museum li-ho said to havebeen excavated in Ho-nan (after Sheng-kao[1940], Pl. 22), note apparent presence of

"rivets" on extended flap. 3: Li-ho said tohave been excavated at Yuan-shui 洹水,An-yang (after Sheng-kao [1940], Pl. 23), note sprue remnant at join of handle and flap. 4: Sekino Takeshi's line-drawing of the li-hoexcavated at Pai-chia-chuang 白家莊,Cheng-chou, in 1955 (after Sundai Shigaku 11[1961]: 14, Fig. 3). 5: Li-ho, B60 B1078,Brundage Collection (after photographs by author), note sprue remnant at join of handle and flap. 6: Li-ho, B60 B53, Brundage Collec-tion (after photographs by author), note sprue remnant, also perforations in handle resulting from core-extension spacers. 7: P'an-lung-

(N.B. Scales of individual items in the illustrations in this survey are not indicated and no attempt has been made to maintain a common scale.)

6

ch'eng 鎜龍城 li-ho (after Hong Kong Exhibi-tion Catalogue 1978, Pl. 13; see also WW 1976.2:34, M2.20). 8: P'ing-ku-hsien 平谷縣,Liu-chia-ho 劉家河 site (after WW 1977.11:3,Pl. 4, 2), Middle Shang context. Dome section is damaged and incomplete. Rubbing of decor band appears below the drawing. 9: Palace Museum (Peking) li-ho (after Ch'i-hsüan [1976],Pl. 7). 10: Po-ku (1110) 15.21. This is theoldest known published reproduction of a li-ho.It is termed here a chia and considered to be

Han period. 11: One of a set of three very elaborately designed li-ho of Late Shang datesaid to have been excavated at An-yang; now in the Nezu Bijutsukan 根津美術館, Tokyo(simplified drawing, after Ch'en Chung-ju 陳仲玉 [1974] 48.4, Fig. 2, 11). 12: A li-hoconsiderably more advanced in form from the Fu Hao 婦好 Tomb (after China Pictorial [1978]1: pages not numbered).

7

bronzes under discussion, belong with the typical inventory of the Ch'i-chia culture.... " No sugges-tion is offered, however, that the assumed metal prototypes were manufactured by smithy techniques.

The most recent published view on the signifi-cance of these and other associated materials is that of Bagley in the course of his detailed study "P'an-lung-ch'eng: A Shang City in Hupei" (Artibus Asiae 39, 3/4 [1977]:165-219) – a timelyand useful translation/collation of the various Chinese reports on this site. But when it comes to his assessment of the significance of some of the materials, an otherwise excellent art historical survey is marred by several misunderstandings of technical matters. Prior consultation with special-ists in our field might have resulted in a less assertive and more cautious presentation. The following passage is representative and convenient-ly contains most of the points which require our attention – these I have taken the liberty to under line:

Yet the extant li-ho vessels, though made bycasting and Middle Shang in date, show un-mistakable dependence on a smithied version, visible in such details as the tubular spout, the projecting ledge between dome and body, and the handle, which is invariably made as a flat horizontal plate projecting from the rim and joined underneath to a strap — an unneces-sarily complex form never seen on any othervessel type.6 Furthermore, metal li-ho ante-dating Middle Shang must have existed at Erh-li-t'ou though none has yet been excavated there, for pottery copies have been found inboth the third and preceding second strata; and these pottery copies retain a featureeliminated from the Middle Shang bronze li-ho, viz., imitation rivets on the top part of thestrap handle where it joins the rim.7 Rivetsin a pottery vessel, at a spot where they wouldbe used on a metal vessel to secure the handle,naturally preclude its being taken as a proto-type for the metal form, and in this case put beyond doubt the existence of a smithied vesselat least as early as the second stratum at Erh-li-t'ou (p. 198).

To what extent may one so boldly assert that the bronze li-ho vessels (eight in all according toBagley; Chang Ch'ang-shou 張長壽 also has eightitems listed in his survey [KKHP 1979.3:275] — line-drawings of twelve items, however, appear in Figure 2) manifest "an unmistakable dependence on a smith-ied version"? He details above the features which he believes prove the point; let us examine the validity of each:

1. Tubular spout: Sekino, too, has placed someemphasis on this functional device. It is under-standable that one may think of metal as the first material in which tubes were fashioned, and that in their earliest form they would be made from beaten sheets of metal. However, one really does not have to search very far to discover that the antiquity of the tube — or, for that matter, even the tubular- shaped spout - is such that in China, at least, a 4700-3000 B.C. dating of assumed metal prototypes would have to be allowable in terms of Bagley's criterion (see Figure 3). The few examples assem-bled in this Figure show also how wide-spread the

imagined tubular-spouted metal prototypes must have been. A little consideration of the material of manufacture, clay, and some understanding of its modeling properties, should be sufficient to confirm the rather obvious fact that it is remarkably simple to make tubular artifacts in clay. Indeed, there are several schematic diagrams throughout the Chinese archaeological literature which illustrate, amongst other things, aspects of the evolution of the pour-ing spout which was achieved fully in Metal Age ceramic contexts, and accordingly without inspiration from metal prototypes. A more extensive study of the pottery data should illustrate more conclusively than the few samples provided here, the questionable value of the tubular spout, or other tubular devices, as a form of proof that the potter was following the relevant feature in a metal prototype.8 The experi-enced investigator should, furthermore, not limit himself to a study of spouts only. If tubular arti-facts have to be explained in terms of a metal proto-type, then surely amongst other exotic pottery structures, the elongated pedestals of pottery tou

豆 it might similarly be taken as further "proof" of such inspiration from metal prototypes? Consider, for instance, the selection in Figure 4, and observe also that there is ample evidence of a developmental process from relatively simple rim-bases to quite elaborate forms; the elongation and elaboration of the body itself, resulting in long tubular structures, might likewise give rise to the opinion that metal prototypes must have inspired such designs.

2. Projecting ledge between dome and body, and thehandle: This ledge Bagley believes, "imitates ahammered type with its cover crimped on and its handle riveted in place" (pers. comm.).9 If thiswere so, it may be considered remarkable that pottery versions antedating examples from Erh-li-t'ou Layer 2 lack positive signs of a representation of this "crimped" ledge. It is in the earlier pottery copies that so pronounced a feature should be found present if these pottery vessels do, in fact, simulate metal prototypes. That the feature is conspicuously absent in Early Shang or pre-Shang ceramic li-hostyle vessels would simply be because metal versions of the kind appearing in Figure 2 had not been castin bronze up to that time.

Now, why does the pronounced ledge appear on the bronze li-ho? And why does this method of handle attachment to the assumed originally forged extension of the ledge occur? The ledge, amongst other evi-dence, simply indicates that the mold-assembly was divided horizontally into two main sections as demon-strated in Figure 5. The upper unit comprised in a 1- piece (2-divisions) assembly formed around a clay model of the dome and spout with the pre-assembly join-line running vertically through the spout in the case of B60.B53 (and so, too, B60.B1078, see further details in the Postscript). The core, A, was produced by "casting" a solid clay impression from the fully assembled molds and whittling it down to the required wall thickness when cast in bronze.10A normal 3-piece assembly with coring provision for the handle, C (in the case of B60.B53 but not B60.B1078) comprised the lower unit. The handle of B60.B1078 has pronounced flashes of a single inter-mold join run-ning vertically along the center under the surface thus showing that mold sections, b1 and b3, continuedto form "extended cores." The band of decor under the handle naturally continues with only a single

88

Fig. 3. Tubular spouts in early Chinese ceramic li-ho and ho style utensils. Theselection presented here is somewhat randomly chosen; the aim is simply to illustrate the wide geographical distribution, aspects of . development, and the high antiquity of ceramic tubular devices. 1: Ta-wen-k'ou 大汶口examples from Periods Five to Ten demonstra-ting the development of the two types of spouted vessels extracted from WW 1978.4.63.Two radiocarbon dates have recently been pub-lished: ZK-317 and ZK-391-O; when calibrated [DFLW] these are 2336 ±134 B.C. and 2024 ±191 B.C., respectively. 2: An embryo version of the spout from Ho-mu-tu 河姆渡, 2nd

layer, Ma-chia-pin 馬家濱 Culture (after KKHP 1978.1:Pl.13.6). Relevant C-14 dates are ZK-46 and BK-75058; when calibrated (DFLW) these are 4749 ±120 B.C. and 3718 ±135 B.C. 3: A ho(T4:420) similar to the Andersson item (K5523) in Fig. 1 from Lung-shan Cultural level, Hsia- wang-kang 下王崗, Hsi-chuan 馬淅川 Ho-nan (after WW 19172.10:18); note absence of"rivets" on upper section of handle. 4: Li-hofrom Period Three burial (M45:25) Ma-pu Gorge

馬壩石峽, Ch'ü-chiang 曲江, Kuangtung(after WW 1978.7:6); the burial M43 of thesame Period has been C-14 dated; when calibrated (DFLW) it is 2863 ±136 B.C. (loc. cit.p. 10; laboratory details not yet published.

9

5: Long-spouted ho from a Lung-shan site,Ning-yang 寧陽, Pao-t'ou 堡頭, Shantung(after Hsin-Chung-kuo [1962], p. 20). 6: Erh-li-t'ou Period Three li-ho (K3:10) of the type which inspired such bronze examples as those in Fig. 2 (after KK 1976.4:260, Fig. 2); radio-carbon dates from comparable levels are ZK-257 and ZK-199, when calibrated [DFLW] these are 1429 ±105 B.C. and 1472 ±141 B.C., respective-ly. 7: Ch'ing-lien-kang 青蓮崗 Culture examples (after Kiangsu [1963], Fig.5.17-19).Radiocarbon dates range from around 4000 B.C. 8: Lung-shan li-ho from Chiang-chai 姜寨,

Lin-t'ung 臨潼, Shenhsi (after KK1975.5:Pl. 2, 3). 9: "Tung-hsia-feng 東下馮Style" pottery li-ho M401:3 from LayerNo. 4 located below an Early Shang level at Hsia-hsien 夏縣, Shanhsi (KK 1980.2:99),excavated along with small bronze artifacts, a piece of malachite, etc. Relevant radio- carbon dates are ZK-382, ZK-397, BK-76042, etc., when calibrated (DFLW) these are 1987 ±174 B.C., 1945 ±91 B.C., 1919 ±91 B.C., respectively, thus resulting in the earliest radiocarbon dated find of metal artifacts in China reported to date.

10

Fig. 4. Elongated pedestals, in pottery touand the evolution of this tubular element from simple ku-beaker forms. The Ta-wen-k'ousequences are extracted from the Table in WW1978.4:62; the Ch'ing-lien-kang examples from the same issue, Fig. 2, p. 49. The single tou is from Lung-shan Culture level, Hsia-wang-kang, Hsi-chuan, Honan (after WW 1972.10:12). If such tubular devices as spouts are to be taken as evidence of pottery imita-tions of metal originals, then one must surely consider the comparable significance of the elongated pedestal?

Ch'ing-lien-kang examples

Ta-wen-k'ou sequence

11

Fig. 5. 1: Section (xl - yl : 120° angle butdrawn as normal 1800 cross-section for conveni-ence) through mold assembly for a bronze li-ho-kettle. Outer molds for the body comprise a 3-piece assembly (b1, b2, and b3) which, curi-ously, embraces a large area of the rear sur-faces of the legs. The inter-leg core, B, was stabilized by core-extension spacers. For the dome-shaped lid and spout a 1 piece (2-divi-sions) mold assembly was employed; extensions of core A at e and f would function as core- extension-spacers and thus stabilize this sec-tion of the assembly. Coring of handle, C, is valid for B60.B53 (see Postscript, Fig: 12,

for further details). Sprue remnants on B60. B53, the Berlin vessel, and the P'an-lung-ch'eng vessel are located vertically along the upper to center part of the handle, f1, whilethat of B60.B1078 is restricted to the flap/ handle junction, f2. The feature supposedlyderiving from a so-called "crimped" ledge was formed either by bevelling the contact surface edges of molds a and b as in 2(i) or bronze flashes seeping between these contact surfaces as shown in 2(iii) were tooled and polished to form an even ledge after the molds were re-moved. The figure as well as Nos. 7 and 8 were drawn by Miss Winifred Mumford.

12

join-line interruption. When assembling the units around the core preparatory to casting — the lower unit being assembled first — provision for slight inaccuracies that might result in mismatching of the upper and lower units was effected by a bevel-ling of the adjacent mold edges to form a flange, 2(i). Without this, a rather unsightly and neces-sarily weak join could occur as shown in 2(ii). Alternatively, it may be observed, leakage of the molten bronze into the interface edges of the two units — whether the edges are bevelled or not — would naturally form flashes of metal. These tooled down to form a narrow flange would result in much the same feature, 2(iii). When we take into con-sideration the necessity, or convenience, for horizontal division of the outer molds, it will be immediately evident that the founders took advan-tage of this division to form the handle in the peculiar manner they have.11 Without such division,a normal crescent-shaped handle would have been constructed in exactly the same manner in such ves-sel types as the chüeh, chia, etc. In the case ofthe Berlin vessel a decorative element was incor-porated in the casting surface of the upper divi-sion; this potential does not seem to have gained the interest of those who designed and cast the other vessels (except the highly decorated item in Po-ku, Figure 2, 10).

According to my reconstruction of the li-homold-assembly it is evident that casting would necessarily have been effected with the assembly in a horizontal position, and thus unlike the normal vertical stance (with the vessel in inverted posi-tion) used later. This idea is purely tentative pending an opportunity to examine the excavated vessels directly. It is, furthermore, over 10 years since I examined the two Brundage vessels.12

3. Imitation "rivets": It is not good scholarlypractice to assign significant terms to assumed technical features unless there is reliable evidence to support one's understanding of the original function of the features. That these protruber-ances are definitely not representations of rivets should become evident upon due study of other Erh- li-t'ou ceramics of the same vintage, and relevant ceramics from other sites (see Figure 6). In the majority of cases they are clearly decorative items, and these later gradually came to be transformed into animal-heads.

Now, let us look at the technical aspects of attaching handles to vessels in smithied artifacts. One point that will already have struck the special-ist rather forcibly is the curiously restricted use of "rivets" in the pottery vessels. When these protruberances appear on handles they are limited to the upper area of the handle terminals and there is nothing to hold the lower handle terminals to the vessel body! Rivets would surely have to be employed at both terminals! Furthermore, it has, no doubt, caught the attention of the reader upon perusal of the examples in Figure 6 that "rivets" placed on the handles, and located so far away from the vessel walls, could hardly be claimed to func-tion as rivets! To claim that they are "at the spot where they would be used on a metal vessel to secure the handle" is incorrect both descriptively and technically. Furthermore, their alleged function is certainly not supported even by the limited evidence

that Bagley has brought to bear on the subject. There does not appear to be any possible support for his claim amongst the considerable corpus of relevant ceramic, or metallurgical, data that he evidently has not investigated.

In Western metallurgical cultures rivets, commonly used and in a large variety of ways. To understand the function of rivets it is necessary to shift one's attention from the Chinese milieu (from earliest times in the history of Chinese metallurgy to circa 300 B.C., during which long period, rivets are quite unknown in the Chinese archaeological scene) to the Middle East, Central and Eastern Europe, Egypt, etc. As our immediate interest is in the use of rivets for the attachment of handles to the bodies of metal vessels, a selec-tion of relevant examples from the Ancient World is presented in Figure 7. These are representative of a large number of examples that could be assembled from just a few books taken at random from appropri-ate sections covering Western metal-working in my personal library. A more systematic and exhaustive search for data on this method of handle attachment in metal-working cultures of the West might well develop into a very interesting research subject in its own right. Here, however, the aim is simply to demonstrate the fact that handles whose upper ter-minals are attached by rivets to the vessel wall will be found also attached by the same means at the lower terminals.13 Also, it is essential toobserve, rounded protruberances do not necessarily imply rivets. The round nodules on the handle in item 5, Figure 7, obviously are decorative; like those on ceramic vessels (in our earlier Figures) which are not only circular in shape, but also hemi-spherical. Accordingly, they can hardly be intendedeven to represent rivets! Rivets are hammered and the ancient metal-worker's intention was that both head and tail of the rivet would flatten and spread to form a mechanical lock. Rivets can, of course, be made decorative; the heads can be pre-formed, and if shaped as a dome, a pyramid, a cone, etc., the shape of one head of the rivet can be preserved during hammering by supporting the pre-shaped head in a depression in a dapping block or punch that fits it in size and shape. This latter technique, which is a comparatively later development, does not concern us here as the riveting in which we are interested is that of the earlier stage which is solely functional and the heads flat.

4. Imitation folded rims (or merely reinforcedrims?): This feature is, as Bagley observes,commonly found amongst bronzes up to the close of Middle Shang after which time it becomes rarer (p. 198). However, its significance in bronze casting practice eludes him because of his preoccupation with smithy work. Admittedly, he has not pushed his interpretation of "folded rims" as strongly as he has in the case of "rivets"; nevertheless, the more casual reader may be inclined to take the inter-pretation as having been proven. With the recent publication of his views in The Great Bronze Age ofChina (Wen Fong, ed., The Metropolitan Museum of Art,New York, 1980, hereafter Bronze Age Catalogue) re-iterated in respect to chüeh (No. 1, p. 74), ting(No. 4, p. 103), ho (No. 5, p. 104), a more extensivedissemination of the "folded rim" theory may be expected.13a Let us get the picture clear at theoutset. The "imitation folded or reinforced rim" in

13

Fig. 6. So-called "rivets" in early Chinese ceramics. 1-6: Various examples from Erh-li-t'ou (after KK 1965.5:219-220, Pl. 4, 9).The placement of the pairs of nodules on the upper terminals of handles so far from point of juncture (items 1-4) obviously has nothing to do with handle attachment to the body — these can hardly be regarded as imitations of

wrought metal features. The association of nodules in item 5 with decor is clear, while nodules in the square vessel (item 6) would never, for instance, be suggested to represent, say, nails (dome-headed, or plain) by anyone conversant with Chinese cabinet work — ob-viously "rivets" would not be used here, or even in the upper leg areas of item 4!

14

7-8: Ma-ch'iao 馬橋 site, Shanghai, Layer 5Liang-chu 良渚 Culture examples, M4:1 and 810:11(after KKHP 1978.1: 116-117). Here, too, it isevident that the nodules are decorative, note the repetition of the nodules on the lid in item 7. The handle in item 8 is reminiscent of metal handles in Fig. 8 (items 1, 3, and 5); if it were actually imitative of a wrought metal original, the placement of "rivets" would be considered quite wrong. 9: A single "rivet" on the handle! (after Kiangsu [1963],Fig. 4, 23). 10: K'o-hsing-chuang 客省莊Period Two Culture (after Feng-hsi [1962],p. 61, Fig. 38). Here, too, the "rivets" are wrongly placed (see relevant examples in Fig. 7

already noted). 11-13: Tung-ma-kou 東馬溝site, Lo-yang (after KK 1978.1:20, Fig. 5, 1-2,7). These Erh-1i-t'ou Period examples (11 and 12) demonstrate well the decorative intent of markings (including "rivets" in 11) on the handles. The handle of item 13 is interesting to compare with that of item 4 above; the center posts give additional strength to the rather flimsy structure of the handle. There are no "rivets" although the angular bend of the up-per section of the handle vaguely brings to mind the bronze li-ho versions in Fig. 2. Thereare many other instances of so-called "rivets" in the early ceramic contexts and all would fit the general category of ornament; none manifest evidence that would indicate the nodules to be imitations of true rivets.

15

Fig. 7. Examples of handles attached to vessel bodies by rivets in early European bronzes (all re-drawn and slightly simplified after Marija Gimbutas Bronze Age Cultures in Central and East-ern Europe [1965], Fig. 27,8; Fig. 11,13-15; Fig. 109; Fig. 110,3). Two main methods are shown; 1 and 3: upper terminal of handle is shaped to fit

against outside wall of vessel and riveted in place.The others have the upper terminal bent around the rim of the vessel and riveted against the insidewall of the vessel. In both cases the lower terminals are always riveted against the outside wall of the vessel.

16

Shang bronzes of the earlier periods has nothing whatsoever to do with metal working. It is simply a representation of the join-line or parting line which earlier formed between mold/core units; and in some cases it is an actual join-line.

In later times (Late Shang) the parting line of core and outer molds generally followed the outsideedge of the rim. Lugs ("handles") were cut directly into the material of the core, thus casting flashesare often found (when not fully tooled) along the outside edges of handles (Figure 8, b and d); and there is no "folded rim" effect. The Middle Shang core/outer mold division was effected along the inside edge of the rim (Figure 8, a and c); lugswere cut directly into the material of the mold,thus casting flashes appear along the inside edges of handles. And the so-called "folded rim" effect results because of the slight variation in level of core and outer mold along the circumferential part-ing line of both. The feature is more easily studied and its significance assessed with reference to ting-cauldrons of which many examples are availableand well described in the reports.14 In the case ofchia, the parting line being located appreciably downthe slope of the inside wall, a distinct circum-ferential ledge resulted. This last feature has apparently given rise to the impression that it derives from a metal-working technique; but one, it should be emphasized, quite alien to the Chinese metallurgical scene prior to Han. The practice of thickening the edges of a metal container, the simple folding of edges (outwards and under) — oreven that of reinforcing the metal edge by folding the metal around wire, or strips of metal - are approaches of high antiquity in the West and com-pletely alien to the Chinese metallurgical scene up to as late as Han times.15

What we see here, therefore, is merely a devel-opmental stage in foundry approaches to mold/core divisions as reflected in vessels with open li-stylelegs (chia, ting, and hsien) and in solid-leg ves-sels such as chüeh and possibly ting (with leaf-blade legs). Other rather significant data is, no doubt, preserved in the P'an-lung-ch'eng and other Middle Shang bronzes of these types which may indi-cate further interesting variations in approaches to mold construction when compared with the bronzes we know so well from Late Shang onwards. But the chances of our obtaining an opportunity to conduct research directly on these vessels is still some-what difficult, while published photographs seldom are taken from the angles we need in order to work on the materials effectively at this distance.

5. Discarding of a technique: An essential element in Bagley's argument, which is perfectly correct insofar as its recognition of the fact that none of the alleged smithied vessel prototypes have as yet been discovered, is the oft-repeated statement to the effect that the early craftsmen discarded com-pletely metal-working technology, and all its prin-ciples, in favor of direct casting in piece-mold assemblies:

On present evidence therefore, smithied vessels existed perhaps as early as the beginning of the second millennium, though traces of them survive only in pottery; and while the presence of the li-ho among the earliest Shang bronzes

conclusively links the beginnings of Shang bronze casting with this elusive earlier stage of metal technology, methods of fabrication other than casting seem to have been discarded in short order by the Shang bronze founders (p. 198).

Let us imagine, for a moment, the situation that would hold if Bagley's theory of the elusive metal-working stage in pre-Middle Shang times is to any extent a viable one. Craftsmen producing smithied metal prototypes of ceramic vessels of the kind presented in Figures 1 and 3 above, would have left their imprint not only in ceramic wares (where the potter might choose to emulate shapes and ornament developed by his metal-working colleagues) but also it would have been preserved in the general tech-nical heritage bequeathed by metal-working artisans to their descendants. Knowledge of the simple metallurgical principle of "annealing," without which the elusive metal prototype vessels could not have been manufactured, would have been preserved in many branches of the metallurgical industry. An-nealing is fundamental to several metal-working techniques and certainly would have been employed in tools and weapons production at the time. Once discovered, or adopted by a metal-using culture, it would be well-nigh impossible for the technique, and its manifold working methods — sinking, raising, riveting, turning, spinning, sheet metal production, etc., — to have disappeared without trace. The products, the tools employed, the work-shops, etc., would show up at least sporadically during the course of archaeological excavations. The idea of a metal-working metallurgical culture being super-seded by a strictly casting-conscious culture and all technical knowledge and physical evidence of the former being completely wiped out so that even so fundamental a metallurgical principle as annealing was successfully proscribed, is, to say the least, incredible.

6. Cold metal working: Bagley makes some verycurious statements in respect to metallurgical matters:

The argument from technological differences is succinctly put by Cheng [Te-k'unJ ... when he maintains that not only did the Chinese not use lost-wax, but they were also ignorant of — or, at any rate never used — practically every other way of working metal. Surely this is unlikely on the face of it; and I trust to have shown in my paper that pottery copies of smithied and riveted vessels prove the exis-tence of a pre-Shang technology based on cold-working sheet metal. This is Professor Laplante's conclusion also ....

Why were the Shang bronze founders so deter-mined to do everything they possibly could by direct casting, and so uninterested in other — often simpler — methods? They knew about cold-hammering (how could they not? In any case the pottery copies prove the point) ... (pers. comm. See note 7, above).

Cold-hammering - beating metal (in particular, bronze) without annealing has very considerable lim-itations. A careful reading of some of the relevant sources listed in Metallurgical Remains, and of theseveral other very useful surveys which have come to

17

Fig. 8. The so-called "folded rim" or "rein-forced rim" commonly found amongst Middle Shang bronze vessels. For ease of diagram preparation, the ting-cauldron (six P'an-lung-ch'eng examples are presented above, after WW1976.2:34, Fig. 36) is used here to demonstrate the variant location of core/mold parting lines. It is solely due to this technical aspect of core and mold production that a circumfer-ential ledge slightly higher (and with smoother surface) than the inside level appears in such vessels as ting, chia, chüeh, etc. Duringmold preparation both the casting space for the handles and the ledge upon which they stand were incised directly into the mold. Where post-cast tooling and polishing was in-differently attended to (a not uncommon char-acteristic of Chinese bronzes) casting flashes, or incompletely tooled remnants of flashes, along the core-side edges of the handles pre-

sent evidence of the approach at the ceramic stage. In later times the ledge came to be partly molded by both the core and the molds but the handles were formed by the incising of the core clay. As a result casting flashes occurred along the mold-side edges of handles. It is seldom that traces remain along the edge of the vessel rim; this is because it has easy access for tooling and polishing after casting. The drawing in C is based on the P'an-lung-ch'eng ting-cauldron exhibited recently in Japan (Bronze Age Catalogue, Pl. 4) and in the New York Exhibition this year. As was perfectly evident upon study through the glass case, the casting flashes continued along the inner edge of the handle and the so-called "folded rim" edge. In the fang-ting-cauldron (No. 11), however, most of such evidence seems to have been eradicated by post-cast tooling and polishing.

Casting flashes

Casting flashes

18

my notice since (see below), would assist those of us who are untrained in the technicalities of the field to understand some of the more fundamental aspects of metallurgy. It is essential, too, that laboratory personnel and metallurgical specialists be consulted frequently during the course of one's research and writing.

Apparently Bagley has confused the process of annealing with the technique of "work-hardening" which involves the cold hammering of metals to induce a hard cutting edge.16 As annealing plays arole in work-hardening it seems strange that he should advocate "cold-hammering" only as the means of shaping such metal prototypes as he assumes were copied by ceramists like those in Figure 1 and 3 above. Then does he realize just what metal-working techniques would be involved to produce these ela-borately designed vessels? The three bulbous legs of the li-ho, for instance, would have had to beindividually made then joined together - some form of soldering or brazing would be involved. The same method of joining would serve equally well to affix the spout to the dome-shaped unit and the latter to the body - no need at all to "crimp"-join the two sections! A little study of such authoritative sources as the late E. Wulf's The Traditional Craftsof Persia (1966), Oppi Untracht's Metal Techniquesfor the Craftsman (1968), and Keith Branigan'sAegean Metalwork of the Earl and Middle Bronze Age1974, to name but a few, would assist laymen students of metallurgy to obtain an appreciation of what metal-working is, what can be done and how it is done, and the problems of interpretation con-fronting the archaeologist.17 Then one must havesome appreciation of ceramic techniques; as it is a popular hobby, nowadays, there are numerous books one may consult for an understanding of the potential of design in clay. The versatility of this material as against metal will show fairly conclu-sively, I believe, the fact that the bronze li-howere nothing other than metal copies of pottery prototypes, while the large variety of exotic pottery-ware from numerous pre-Metal Age sites was designed by potters exercising a greatly creative spirit and in full control and understanding of the medium in which they worked.

Here then we approach the basic problem — one to do with the proverbial cart and horse. Which comes first: the imitating of pottery shapes by bronze founders, or the adoption of bronze shapes by the potters? We may be certain that copying from one direction would have commenced the very moment the first Chinese founders attempted to cast con-tainers; the first founders obviously would seek to duplicate, to the best of their ability, the well established shapes of ceramic vessels current at that time. Then as bronze shapes took on special characteristics these would sometimes have been adopted fully or partially by the potters. Later, as bronze vessels underwent further changes in shapes and designs, further copying would have been done in ceramic forms. To what extent there may have been a ping-pong effect - i.e., bronze vessels developed to a certain stage which were then copied by the potters, with these ceramic versions then undergoing changes which in turn may have inspired later bronze founders, and so on - is not presently clear to me. My impression is that the story of pottery imitations over the Shang and Chou periods may be found simply to be one of continued copying

on the part of the potters. The bronze shapes and designs after a formative period of imitation and experiment continued along the lines of development and change which were essentially independent of the ceramic industry. This is a subject worthy of exten-sive study and it would seem to me that the most effective way to conduct the basic research would be to work backwards, commencing, say, from Han times. Working thus from the known to the lesser known, thence towards the unknown, the exercise might well result in the establishment of viable and accept-able methods of extrapolation into the Early Shang and Late Lung-shan ceramic contexts. This might even shed valuable light on the beginnings of the casting of containers in bronze.

In the concluding paragraphs of her survey of the relationship between the painted pottery and Lung-shan Cultures (see note 1, above) Louisa Huber arrives at the opinion amongst other considerations that the major difference between Chung-yüan 中原,and Shantung pottery traditions in the Late Lung-shan period would appear to be this: "the most distinctive of the Chung-yüan vessels do not belong strictly within any ceramic tradition ... namely the kuei andits descendants, the li-ho and ho.... These vessels,as has long been realized, are clearly metallic in character ... imitation rivets on the apron of the handles, the broad thin handles ... flat bottoms (when not raised from the ground by li-legs) ... rela tively thin-walled, all mark them as copies of wrought-metal vessels that also served as the direct models for the earliest cast bronze vessels" (pp. 32- 33).

Several of the preceding "metallic" character-istics have already been dealt with. However, the assumed relationship of "broad thin handles," "thin-walled" vessels, and "flat bottoms" of solid leg vessels requires some comment. It is very much to be doubted, I think, that a tendency towards thin-ness of the walls, handles, etc., in pottery vessels can be regarded as evidence of copying from wrought metal vessels which similarly are assumed to have been of thin wall construction. After all, if a pottery handle is to be reasonably strong it must be either narrow and thick, or if thin it has to be appreciably wide in order to compensate for the weakness of the thin ceramic wall. Flat bottomed vessels of various shapes are commonly found in Yang-shao ware and sometimes with solid legs; the development of legged vessels with flat bottoms constitutes a ceramic tradition entirely different from that of the li in its various manifestations.Neither form of legged vessel owes its inspiration to metal prototypes - no matter how hypothetical — while the alleged existence of wrought metal proto-types is based upon an insecure knowledge of metal-lurgy as well as the way bronze vessels of the types under consideration might perhaps be worked into shape by a smithy, or (more simply) cast in a foun-dry.

We must recall here, too, that many of the earliest cast bronze containers have very thin walls; the tantalizing chüeh (item 1 in the Bronze AgeCatalogue) was the subject of much discussion andspeculation between Wang Chia-pao and me in Taipei last August. To cast metal so thinly and often so successfully raises a host of problems which to my mind may not be possible of solution until such time

19

as we attempt to duplicate the process of casting of comparable items under laboratory conditions.18And foremost amongst the questions that arise is this: are there known thin-walled castings in other Ancient World cultures of the same time (say, to-wards the end of the 2nd millennium B.C.) or earlier, of comparable standard and complexity in design?

POSTSCRIPT

A year or more has passed since the preceding survey was written and typed up into what I expected would be practically final form — i.e., the version which was duplicated and distributed to participants in the Berkeley Workshop "Ancient Chinese Inscrip-tions and Bronzes: New Discoveries" (7-9 June, 1980) and read together with another paper "Some Observations on Metal Winning, and the 'Societal Requirements' of Early Metal Production in China" in the morning session, 8 June. Most of the present year, however, has been spent in travel and research in China, Taiwan, Japan, and the USA, and partici-pation in Conferences, Symposiums, or Workshops in New York, Berkeley, San Francisco, and Taipei. Thus opportunities to obtain further data relating to past and current research projects and those planned and in progress for the near future were many, whilst numerous meetings with colleagues in practi-cally every place visited has resulted in stimulus of a high order in nearly all facets of my researches in the various fields in which I am interested or active. Accordingly, it has been found necessary to revise some sections of the survey and to ela-borate considerably my earlier technical notes on the two Brundage li-ho (kuei) which I examined indetail while in San Francisco, the Berlin li-ho(two copies of Chinesische Bronzen, Berlin, 1928came to my attention in the Fogg Art Museum Library), and the P'an-lung-ch'eng li-ho which was on displayin the "Great Bronze Age of China Exhibition" in the Metropolitan Museum, and to incorporate notes on or references to, a number of other more or less relevant materials in the Exhibition. Several im-portant studies recently appearing in the Chinese archaeological journals have also been examined and due attention given to them; and some mishaps which appear in the Berkeley Workshop "hand-out" have been duly corrected. In addition it has been pos-sible to incorporate a few scattered notes and com-ments touching upon sections of the Bronze AgeCatalogue.

To deal with the relevant passages more fully, it would require either a considerable amount of . recasting of the survey and that would mean practi-cally re-writing it, or the writing of a review of the Bronze Age Catalogue. However, the lattercourse does not appeal to me partly because of more urgent work awaiting attention, and partly because the seal of authority is so firmly impressed upon this publication, together with the Museum's approval of the contents in question (demonstrated other-wise in the wording of the captions accompanying the relevant artifacts on display). This would make it well-nigh impossible for mere scholars to seek to correct misconceptions propagated thus so effectively amongst so wide a section of the gen-eral public.19 Here, therefore, I shall simply citeseveral representative passages and indicate which statements of fact therein are incorrect or open to question (underline plus asterisk*). Those state-ments which have been dealt with earlier in the survey are included but generally require little further comment; those which have not received

attention have a numeral typed alongside the asterisk*l and a corresponding note appears below the cited passage. To avoid the confusion engendered by the Quixotic p'in-yin 拼音 romanization usedin the Bronze Age Catalogue and throughout theExhibition itself, I have substituted Wade-Giles:

Except for the spout, the vessel seems to have been cast in a single pour. The spoutwas evidently added in a second operation;*la teardrop-shaped puddle at its base belongs to this later pour.*2

In many respects the [li-ho] shape is uncharac-teristic of cast bronzes; its domed cover, for instance, which makes removal of the coreafter casting very inconvenient,*3 is afeature that would hardly be invented in a founder's workshop. Thus while the present vessel was made entirely by casting, it offers tantalizing clues to an earlier stage of metal technology that relied less exclusively on the casting process. The lobed lower part imitates an older pottery shape, as already noted. The cover, on the other hand, imitates a wrought metal object so accurately that it can be used to illustrate the process of manu-facture of its prototype. The domed shape is a typical product of the smiths' hammering technique, and the tubular spout rolled fromsheet metal* is equally familiar in wrought-metal traditions. The cover would have beenjoined to the vessel proper by flaring the the edges of both pieces and folding one over the other, a technique known as crimping; the crimped join is meticulously reproduced here.*4Lastly, the handle of the present [li-ho] hasa form typical of wrought-metal vessels,* notof cast bronzes. It imitates a composite structure whose horizontal upper section was hammered out from the rim of the vessel then riveted to the separate strap below. (The lower end of the handle which need not be sofirmly joined to the vessel,*5 would typicallyhave been soldered on or secured by an adhesive.The spout would probably have been attached in the same way.*6 In China, only the [li-ho]vessel has a handle of this form;* the simplestrap handles of other early bronzes invariablylack the unnecessary horizontal upper art*(nos. 1, 6, 15, 16). Among the earliest Shang bronzes the [li-ho] thus stands apart, beinguniquely close to some undiscovered wrought-metal type of still greater antiquity. Potterycopies of such a prototype, duplicating the domed lid made into a face and the composite handle of the present vessel are known from a few sites that long predate the Shang period Bagley, 1977, pp. 196-98).*7 These potteryimitations of metal vessels reveal the exis-tence in China of a developed metal technology long before the earliest bronze vessel known at present e.g. no. 1).*8

A bronze [li-ho] slightly more advancedthan the present example was found in the tomb of Fu Hao at Anyang (chap. 4); it must already have been an antique at the time it was buried (China Pictorial 1978.1:25; Bronze Age Catalogue

20

p. 104). *8a

*1 If a two-stage process, the lid would be cast on to the appurtenance; in this case, the spout would be cast first, inserted into the mold-assembly for the lid, and second-pour metal forming the lid would lock the pre-cast spout in place, or possibly form a "metallurgical join." However, neither the P'an-lung-ch'eng li-ho (which Bagley isdescribing here) nor the two Brundage vessels show the slightest evidence that might suggest a two-stage process. And, there is no ambiguity of evi-dence to warrant such speculation (see technical notes below).

*2 Absolutely incorrect. The "teardrop-shaped puddle" is not only an integral part of the single pour, forming the lid + spout + vessel + handle complex, but also it provides important evidence of vertical mold-division (see technical notes below) of the domed lid + spout section.

*3 Removal of the core poses problems in anyvessel. The hard-baked clay has to be forcibly removed with hammer and chisel-like tools (see Wan Chia-pao's fifth volume in the Ku-ch'i-wu yen-chiu chuan-k'an 古器物研究專利 series; curi-ously Bagley lists only the third volume in his bibliography!). It would be just as "inconvenient" in the case of the fragile chüeh (#1), the narrowneck yu (#9), the elephant hsi-tsu (#24), thehorse hsi-tsun belonging to the Lai (or, Li) set(#59 & 60, but not included in the Exhibition), etc., to dig out the cores in each. The openings in the last two are slightly smaller than those in the li-ho!

*4 As observed earlier in the survey the term "crimp" has been incorrectly used. However, the process of folding the flared edges in the reverse direction (as would be done here) would naturally result in a "crimpled" effect. Possibly Bagley is confused on this point. Be that as it may, it is a fact that the reverse folding of the flared edge would result in numerous pleats, and these, too, would have to be reproduced in the cast version (no less than in pottery versions) if one is to assert that "the crimped join is meticulously re-produced." And further, Bagley is now essentially arguing here that the P'an-lung-ch'eng li-ho copiesfeatures in extant metal-worked prototypes! Surelyconfusion of this order should have been questioned at the editorial level or by readers?

*5 A handle joined firmly to the vessel at its upper terminal and less firmly at its lower terminal? Doubtful!

*6 A handle joined by riveting at the upper termi-nal and by solder at the lower terminal? Very doubtful, indeed. And especially so in a metallur-gical culture which did not use solder before Ch'un-ch'iu times! And as to using an "adhesive!"

*7 So far as pottery "copies" duplicating the "composite handle of the present vessel" are con-cerned, there are none. See Figures 3 and 6, which include most of the relevant pottery items Bagley refers to in his P'an-lung-ch'eng paper; also Figure 1 for the Andersson example; there is no example amongst these with a representation of

the circumferential "crimped" ledge let alone a horizontal extension of it with a "strap handle" attached by "rivets" to it. As to these "pottery copies" ... "duplicating the domed lid made into a face," some stretching of the imagination is re-quired to accept this interpretation. None of the pottery dome structures could be claimed to have the appearance of an animal, or human, face except in a rather superficial way — so much so that they can hardly be claimed to have copied such features assumed to have been present in the hypothetical metal-worked dome lid. But if we shift our attention to the bronze founders' renderings of the li-ho, which are so obviously inspired by pottery proto-types, more precise interpretations of the dome structural features will be found possible. Viewed from the handle side — notwithstanding the redundancy of the spout – the appearance of a face is strongly suggested mainly because of the changed shape of the mouth maintained in each vessel. The Berlin vessel with the addition of a nose clinches this impression. If viewed from the front (the spout side) a twin-domed effect in each bronze li-ho nowreplaces the earlier smooth single-dome structure characteristic of the pottery li-ho. With theerect spout situated as it is, an unmistakable phallic symbolism results.

I would comment further here on the complete lack of illustration of the various pottery li-hopurportedly preserving the characteristics of the hypothetical metal-worked prototypes in Bagley's P'an-lung-ch'eng paper. With these crucial illus-trative materials placed right before the reader, as in the present survey, a clearer and more accu-rate picture of what the author is actually discus-sing (and describing) results.

*8 Regarding the excavation of Chüeh No. 1 and itsdating see notes below next passage.

*8a A line-drawing based on the China Pictorialillustration has been added to Figure 2. It would, I think, be somewhat open to question to regard this vessel "as an antique at the time it was buried." It is no less Late Shang in its design and execution than the well known set of three li-ho in the Nezu Bijutsukan which are supposed to have been excavated at An-yang. And to describe it as being "slightly more advanced" [my italics]than the P'an-lung-ch'eng li-ho is no less open toquestion. The profile of the legs (each cylindrical and solid cast at the base) differs markedly; so, too, the distinctive neck; the "apron" (or "flap") and "strap" type handle have been supplanted by an animal-head handle which is U-shaped in cross section and core-cast; the "crimped" ledge appears to be more exaggerated in the China Pictorialphotograph while the dome lid is of lower relief and possibly without division into lobes on the spout side (this last point requires confirmation). It is, of course, possible that we see here aspects of typological development. I believe it is advis-able to think in terms of sequences along the lines of that proposed by Ch'en Chung-ju ("Ch'ing-t'ung ho-hsing-ch'i ti yen-chiu" 青銅盉形器的研究, Ta-lu tsa-chih 大陸雜誌 [1974] 48.4, Fig. 2). Typologically (in design) and technically (in manufacture) it is considerably in advance of the several li-ho (excluding the Nezu items) in our

21

Figure 2.

There are occasionally found items amongst tomb furnishings which were, in all probability, antiques at the time of burial. One example which immediately comes to mind is the bronze p'ou

瓿 of Western Chou date in a Western Han tomb(wooden burial chamber 木槨墓) excavated atLao-fu-shan 老福山, in the northern suburbs ofNan-ch'ang 南昌, Kianghsi (KK 1965.6: 268-272,300). Such examples, however, are rare.

This fastidiously shaped vessel [Chüeh No. 1]represents a considerable artistic advance beyond its ungainly predecessor from the samestratum at [Er-li-t'ou] Fig. 15....*9 Be-sides the three outer sections, the mold assembly included a semicircular or D-shaped solid core occupying the space between the handle and the vessel wall. The two slots inthe handle, a feature known from a few otherearly bronzes, must have been left by pegsthat served to hold this core in place duringcasting ....*10 The present [chüeh] is the ear-liest decorated Chinese bronze so far known. ... Although the entire [chüeh] was made inone casting, its rim seems deliberately toimitate the inward-folded edge that might be expected in a vessel made from sheet metal.*This feature, which also can be seen on the [chüeh] of figure 15, may indicate that the[Er-li-t'ou] bronzes, the earliest known, draw on a still more archaic stage of metalworking, when objects were formed from sheet metal or hammered into shape rather than cast (see entry nos. 4, 5).

Two stubby posts on the rim at the base of the spout, triangular in section, may also have originated in some step of fabri-cation.... The suggestion that they are remnants of pouring inlets for the molten bronze is unconvincing, their location being ill-chosen for that purpose....*11 It is puz-zling that the earliest [chüeh] known so far, that of figure 15, lacks any trace of the posts;*12 (Bronze Age Catalogue, pp. 74-75).

*9 Provenance data is incorrect. Under the next Catalogue entry, there are further details relating to one of the jade blades obtained from the same excavation — " ... found with nos. 1 and 3 in the stratum of the palace remains at [Erh-li-t'ou]" — which illustrate the nature of the error. Chüeh No. 1 (and the jade blades) are reported to have been found in the summer of 1975 near the Erh-li-t'ou site and about 20m south of the Ssu-chüeh-lou ta- tui-pu 四⻆樓大隊部 (see Figure 10). The ex-cavators believe these and some other items derive from a burial and that other burials might be located nearby (see KK 1978.4:270 - this report is alsocited by Bagley). The "earliest known [chüeh]" inFig. 15 (of the Bronze Age Catalogue) which is re-produced from the English language periodical China Pictorial (1977.2:22 acc. caption) but is reportedelsewhere in some detail (KK 1975.5:304, 305) and,as we shall see, far more reliably than in the popu-lar media which Bagley has consulted. The archaeo-

logical report mentions two things: this chüeh(VIII T22③:6) was recovered from sector 8 in thespring of 1973 from a location about 150m northwest of the Early Shang Palace foundations. Found along with the chüeh were several jade artifacts — allfully described (p. 306). Few other details relat-ing to these trial diggings are reported. It is observed, however, that the stratum in which the chüeh was recovered when compared with the samestratum in other places, and the artifacts found therein, show it to belong to Erh-li-t'ou Period Three. Now, the report goes on to relate that in October, 1974, the same team recovered a chüeh simi-lar to the present one (VIII T22③:6) in an ashysoil layer in Sector 4 in the southern side of the Early Shang Palace foundations. In KK 1976.4, thisis again referred to (p. 260), and two further chüehexcavated on this occasion are reported and illus-trated — K3.4 deriving from an Erh-li-t'ou Period Three burial and the other chüeh of uncertain prov-enance (採集). The October, 1974 chüeh has notyet, apparently, been reproduced in illustration.

A brief study of these data shows that based on the dates of excavation alone Chüeh No. 1 in theBronze Age Catalogue has nothing to do with theOctober, 1974 item with which it has somehow been confused! Consultation of the site map, too, would have helped clarify the position. So far as ChüehNo. 1 is concerned, it may be stated that (a) it was not found in the same layer as VIII T22③:6,and the report gives no indication as to its archae-ological associations; or (b) possibly it is the "earliest decorated Chinese bronze so far known," but its lack of secure archaeological context would require a more cautiously worded statement. Never-theless, Bagley is possibly correct in believing it to be the "earliest decorated Chinese bronze so far known" notwithstanding the fact that the relevant report does not say at all in what layer it was found.

*10 If a D-shaped core were used, the adjacent mold sections forming the body and the outside of the handle would have held the core in place without need for "pegs." In any case use of "pegs" shaped thus would be rather unlikely. I cannot comment specifically on the method used for casting the handle as direct inspection (i.e., the vessel being removed from the case for effective study) would be necessary. However, other than the handle problem Bagley has reconstructed the casting method reason-ably well in the section omitted from the passage cited here. Tentatively I would expect to find the slotted handles (see the K3.4 vessel in Figure 10) to be simply a decorative feature.

*11 Naturally I cannot side-step the issue of the knob-stems in chia and chüeh but I do not wish toget involved too deeply in the subject in this sur-vey. However, the statement that (as pouring inlets) "their location [is] ill-chosen for that purpose" is no less pretentious than the grandiloquent assess-ment that the function suggested for them is "un-convincing." Orotundity is no substitute for schol-arship.

*12 One should not jump to conclusions. If no knob-stems ("posts") are evident, it does not necessarily mean that originally there were no knob-stems. Not infrequently in the reports there is reference to chüeh and chia being excavated with these members

22

Fig. 9. A bronze li-ho in the Hsi-Ch'ingku-chien (32.16a) with comparable potteryprototypes from Ta-wen-k'ou, Shantung Lung-shan and Honan Lung-shan Cultural remains (after Chao Wang-p'ing, see note 4 for de-tails).

As a matter of general interest a copy of a recently discovered painting by the well known artist Kao Feng-han 高鳯翰 (1683-1748)is incorporated. It depicts a Lung-shan type of pottery kuei stated in the colophon to havebeen unearthed near Chieh-tzu-ch'eng 介子城where ancient pottery has often come to light. A similar grey earthenware vessel is described

in the same article (Liu Tun-yüan 劉敦愿,"Ken-chü yi-chang-hua hsün-chao-tao ti Lung-shan wen-hua yi-chih" 根據一張古畫尋找到的龍山文化遺址, Wen-shih-che 文史哲 1963.2: 37-41) with the same colophon wording plus the date (1745) incised in it. As the vessel is obviously a facsimile (p. 37) and doubtless patterned after the original exca-vated piece, it would seem to me to suggest that the background of the Hsi-Ch'ing ku-chien bronze vessel may have much in common in view of the fact that no such bronze vessels have yet been excavated. In other words it, too, may be a manufacture of recent centuries.

1-4 Ta-wen-k'ou Pottery 5-13 Shantung Lungshan Pottery

Honan Lungshan Pottery

23

Fig. 10. 1: Map of the Erh-li-t'ou and Ssu-chueh-lou area. Chüeh No. 1 and the jadeblades are reported to have been found in the vicinity of area B. The Early Shang palace foundation remains are located in area A. 2: A line drawing of Chüeh No. 1 and two of thejade blades (after KK 1978.4:270). 3: Chüeh,K3.4, excavated from an Erh-li-t'ou Period Three burial; it has slotted handles and lacks knob-stems. The chüeh on the left is not fullyprovenanced (after KK 1976.4:260).

Fig. 11. 1: A sketch based upon a photograph of chüeh VIII T22 3 :6 and a line-drawing ofsame (after KK 1975.5:305, Pl. 9, 2). Thevessel has been reconstructed along the spout edge but only the knob-stem bases are indicated — the reconstructors not knowing the exact shape have not added these features. 2: Bagley's Fig. 15 version from China Pictorial where theoriginal nature of the corroded and damaged edge is strongly suggestive of breakage or loss of knob-stems (sketch based upon the ChinaPictorial photograph).

24

Fig. 12, 1 and 2: Details evident in the inter-leg areas of 860.853 and 860.81078 respectively. 3 - 5: The spout and dome-shaped "lid," rear and front views of B60.B53, and the rear view of B60.B1078. 6 and 7: Details of the handles of B60.B53 and B60.B1078. Sketches taken directly from author's field-notes; scales and shapes are rough approximations only. (These drawings by Janice Eklund are based directly upon the sketches submitted to us by the author. The sketches were redrawn — without consultation with the author — because of the demands of our format.)

original repairs with sprue remnants

second pour fillings of core-extension spacers

recent repair using plumbers solder

1

2

3

4

5

vertical join traces visible

second pour repair

second pour repair made before dismantling of outer moulds

fossilized straw (?) matting

25

sprue ridge indicating sprue entry area

perforation intended

perforation resulting from entrapped air and gases

6

join traces along line of handle core

shifting of upper (lid) and lower (body) mould assemblies. 660.653 has same feature.

single join trace - no handle core

repair patch-second pour metal

7

sprue entry in this area

join trace

26

missing as a result of breakage during excavation, corrosion, etc. Knowing this situation, one would be expected to exercise caution. Further, excavated vessels are often repaired and some degree of recon-struction effected but usually not to the extent of adding, say, a new pair of knob-stems when the nature of originals is unknown.

Now in the case of Bagley's Fig. 15 chüeh (viz.VIII T22③ :6) a little research shows that thevessel ha undergone reconstruction (KK 1976.4:260);both the photograph (Pl. 9.2) and the line-drawing (Fig. 4.3) in KK 1975.5 are based on the recon-structed version. The China Pictorial photograph repro-duced by Bagley in his Fig. 15, however, shows the vessel before repair. These materials are broughttogether in our Figure 11. It appears evident that originally there were knob-stems on this vessel. The reports, unfortunately, say nothing one way or the other. The "puzzle," or problem, confronting us is actually to determine whether the evidence we are able to discover, upon due investigation, may be taken to indicate that there were originally knob-stems, or not. There is no ground, so far as I am aware, to allow one to state empirically: "It is puzzling that" the vessel "lacks any trace of theposts." Researchers must at all times keep theireyes open for trivial-appearing data of this kind in the text of the reports if they are to get somewhere near the truth of the matter. The archae-ological reports are seldom as detailed, or as comprehensive, as we might wish them to be. I think most researchers are aware of the situation and know the difficulties confronting the Chinese archae-ologists and the journals through which they publish their reports with so vast an amount of material coming to light. To get as much data as possible into the limited publishing space available, reporters often have no choice but to short-circuit details and to be selective. Accordingly, those of us con-ducting research based on the reports and any other sources of information that may come our way must exercise extreme care in our publications. Even so, mishaps may sometimes occur. Those who do not re-search deeply, or who merely work from illustrations and pay minimal heed to the texts of the reports, should try to avoid empirical statements or generali-zations.

In case there may be queries as to the possi-bility that, notwithstanding the breakage evident in VIII T22③:6, the vessel may actually never havehad posts — the breakage being simply coincidental, I will give my response right now. First, the sug-gested reading of the evidence as given above — assuming that there is no other relevant evidence hidden away elsewhere in the reports — is as far as we can go towards a reasonably acceptable interpre-tation. Second, if one wishes to posit the possi-bility that chüeh cast in antiquity may sometimeshave been made without "posts," there is no need tospeculate fruitlessly in regard to the above chüehat all. Simply check the available evidence con-cerning K3:4! So far as I can judge from the avail-able data, this vessel, which derives from an Erh-li-t'ou Period Three context, seems almost certainly to have been cast without knob-stems. Being so rare an exception to the general rule, I would be inclined at present merely to remark upon this curious state of affairs and leave further discussion until such time as the vessel can be examined directly, or other

chüeh or chia without knob-stems are excavated undercontrolled conditions.

As this is a second chüeh excavated from anErh-li-t'ou Period Three context, and reported in publication nearly 12 months before the China Pic-torial vessel, Bagley's claim that the latter is the "earliest" would seem to require correction.

Numbers of other passages might be cited each with much the same number of underlined statements and with the necessary corrections noted below; to continue thus is tedious. The manuscript should have been submitted to competent readers beforefinal acceptance for publication.

Comments on Figure 12

Re-examination of the two Brundage li-ho (June,1980) has revealed several details of interest which it is convenient to record here. In Figure 5, attention is drawn to the unusual location of the sprue which in the case of B60.B1078 varies slightly from B60.B53. The latter, however, accords with the P'an-lung-ch'eng and the Berlin Museum examples. On this basis it would appear evident that these four vessels (and possibly the other li-ho) were cast ina horizontal position with the handles in upright position. Study of the inter-leg areas of the Brundage vessels (Figure 12, 1 and 2) shows that the outer mold enclosure of the conical-shaped ends of the legs, and the nature of the inter-leg core unit, preclude sprue location in the expected places (at the ends of the legs, or in the small flat triangular area between the inner bases of the legs). We may, accordingly, be certain that the two vessels would not have been cast in the normal inverted position.

Core-extension-spacers connecting with the inter-leg core held the three leg cores and the inter-leg core firmly in position. With the removal of the core clay after casting, the three perforations left in the rear surfaces of each of the legs of each vessel were filled in with molten bronze. The tech-nique of filling with second-pour metal such incom-pletely cast areas of the bronzes is commonly found in Chinese bronzes from the Shang period onwards. In B60.B53 there are several such filled in areas to be noted, other than the perforations left by the core-extension spacers (Figure 12, 1) and two of these in the legs have fallen away and have been soldered back in place with plumbers' solder in re-cent times. Between the flap section of the handle and the mouth of the dome "lid" there is a large second-pour repair which appears to have been effected before the outer molds had been fully dismantled (Figure 12, 3), and a sprue ridge remaining on the inside bears witness to the direction of pour. Sim-ilar small repairs with sprue ridge remnants may be noted at the base of one of the legs.

Join-traces running vertically along the under-side of the spout of B60.B53 and continuing down the "tear-drop-shaped puddle" illustrate the nature of mold division for the spout and dome-shaped unit. Much the same evidence is preserved in B60.B1078. In this vessel the high relief modelling of a nose is of special interest. Fossilized remnants of straw matting still adhering to the dome surface alongside the mouth are representative of a class of criteria that may be considered to provide positive proof of

27

authenticity.

Casting division of the handles in the two vessels is quite different as may be observed from the details and associated captions in Figure 12, 6 and 7.

Details such as those noted above, and earlier in this survey, may be expected to allow interpre-tations of some significance when the whole series of presently available, or extant, li-ho are eachexamined intensively. Variations in foundry ap-proaches, for instance, may have geographical or chronological significance. These studied in rela-tion to vessel shape and decor may further enlighten the researcher seeking sequential evidence for a group of this kind. Now with no less than four items properly provenanced a better understanding of those which are less well attested, as well as those lacking provenance data, should be possible. It is with thoughts along such lines that research of value may be undertaken. Speculative avenues producing sensational conclusions that fall apart upon re-examination of the data provided in support of these conclusions (as well as such data which somehow was omitted, or not thoroughly or effectively investigated) are best left alone. After all, few researchers are going to be so fortunate as to dis-cover fundamental truths so effectively hidden from their contemporaries and predecessors specializing in a given field which spectacular discoveries will require a complete readjustment of ideas. If one's preliminary investigations suggest such a conclusion to be feasible, it is best to treat it in the early stages as a danger signal, and go over the ground repeatedly, intensively, and extensively.

FOOTNOTES

1. In this short appraisal of the nature and the validity of the evidence that has been brought to bear on the subject, I shall not refer to the second paper (Louisa G. Fitzgerald Huber, "The Relationship Between the Painted Pottery and Lung-shan Cultures") until later in the discussion. It would be more convenient to assess the arguments relating mainly to technical aspects of the relevant processes of manufacture first, then gradually shift our attention to wider issues.

2. This statement, of course, takes into account the limited tooling and polishing of bronzes following their release from the molds. Abrasion traces re-sulting from the application of implements fashioned from stones with various abrasive characteristics are often easily discerned in areas where such post-cast manipulations would be required. Eradication of casting fins would be the major purpose of such tooling and polishing. Thin spicules of metal along the edges of decoration which resulted from fine incisions made in the casting surface of clay molds, e.g., the outline cutting of lei-wen filling withremoval of surplus clay between the cuts (cf. R.J. Gettens, The Freer Chinese Bronzes, Technical Studies[1979], p. 118, Fig. 147) would likewise need to be eradicated by simple polishing procedures. Activi-ties of this kind are to be classed only as "finish-ing"; they are not a form of metal-working. The alleged presence of primitive hammered implements in the Huang-niang-niang-t'ai 皇娘娘台 site, Wu-wei

武威, Kansu (KKHP 1960.2:60) is rejected; see

Kao Lin-sheng 高林生, KK 1962.2: 99. Interesting -ly, in the latest reported diggings at this site (4th season, KKHP 1978.4:436) one of two knives ex-cavated (T18:6) is described as "hammered." This, too, must be questioned pending laboratory examina-tion of the item.

3. This vessel, I have been given to understand, is no longer in the Museum of Far Eastern Antiquities, and quite some time ago was "returned to Peking." I was in Stockholm for a few days following the Moes-gard Symposium (21-25 November 1978) and naturally, amongst other requests, asked to see this ceramic piece.

4. Although An gives no bibliographical data, it seems certain enough that he has in fact followed Bylin-Althin's article. It is interesting to note also his reference to an example of the li-ho-kettle— he uses the character 鬹 kuei for this vesseltype — in the Hsi-Ch'ing ku-chien 西清古鑑(see 32:16a) which is a bronze version of the Lung-shan ceramic types (see our Figure 9). The authen-ticity of the Hsi-Ch'ing vessel would naturally be a matter for consideration; however, if vessels of this kind should come to light and can be dated to Middle Shang or earlier, they would usefully demon-strate the propensity of the early founders to copy extant ceramic shapes. Early this year a short sur-vey dealing with this li-ho (or kuei) type by ChaoWang-p'ing 邵望平,"T'ung-kuei ti k'ai-shih"

銅鬹的啓示 (WW 1980.2:86-89) appeared. He dis-cusses the possibility that excavated pottery kueifrom such Shantung Lung-shan Culture sites as Ta-wen-k'ou 大汶口,might be connected with theEastern Yi-barbarians 東夷. (on this point see alsoLiu Hsin-chien 劉心健 and Fan Hua 范華,"Ts'ungt'ao-kuei t'an-ch'i" 從陶鬹談起, Ku-kungyuan-k'an 故宮院刋 1979.2:45, 87), while theHsi-ch'ing bronze li-ho may be taken to indicatethat bronze founding (青銅冶鑄業) was alreadypracticed circa 2100-1500 B.C. in the area. Thislatter point is, of course, open to question not only because of the lack of data on the provenance of the Hsi-ch'ing vessel but also because the inscriptionin the vessel would suggest a date probably no earlier than Late Shang. However, Chao is on the right track in believing that such bronze li-ho are renderingsof extant pottery originals. This propensity, in the early stages of Chinese metallurgy, will be dis-cussed at greater length later in the present paper.

5. I regret the necessity of having to point out here the somewhat careless reading of this footnote by Bagley in his note 46, p. 198, in his survey (1977) shortly to be cited. He writes: "Much the same position [against the wrought metal hypothesis] is taken by Noel Barnard, even though (following Sekino Takeshi) he notes the smithied character of the li-hotype (... p. 8, n. 9)." This statement of my opinion is not accurate and is misleading. I most decidedly have not followed Sekino or "noted the smithied character of the li-ho type. "On the contrary, I have stated in plain English: "They are quite

28

definite1y not wrought bronze work, in whole or in part." Then later in my footnote in regard to an example in the Berlin Museum which I have not per-sonally examined (this point is stressed) but which Sekino has described in detail, I have stated: "Certain aspects of the Berlin piece understandablywould suggest smithy work — the neck and handle inparticular." Read in the full context of my anno-tation it should be clear, however, that this sentence is merely my appraisal of why Sekino con-cluded that the three li-ho bronzes he studiedpresented evidence of metal-working.

6. There comes to mind immediately, however, the Middle Chou yi-ewer 匜 in the National Gallery ofVictoria Collection with precisely this feature; see, for example, Hsi-Ch'ing (1751/52), 32:24a;Ku-kung (1958), A.223; T'ung-k'ao 2 (1941), Pl. 861;Hai-wai (1935), Pl. 116, for similar pieces with the same features attending handles. The National Gallery of Victoria yi-ewer has remnant clay corein the legs which has been subjected to thermo-luminescence tests and demonstrated to be datable circa 700 B.C. (see A.J. Mortlock, "The Thermo-luminescence Dating of Ancient Chinese Bronzes" in Ancient Chinese Bronzes and Southeast Asian Metal and Other Archaeological Artifacts [1976, NoelBarnard, Editor], p. 241, Table 2, item 5; also my forthcoming paper: "A Yi-ewer of Middle Chou Stylein the Collection of the-National Gallery of Vic-toria" to appear in The Artifact, the Archaeologi-cal Society of Victoria). Amongst properly attest-ed bronze vessels there is at least one example which was unearthed in 1960 at T'ao-wu-chen 陶吳鎮, Chiang-ning 江寧, Kiangsu (see Kiangsu[1963], Pl. 81). In the same publication (Pl. 92) a rather curious p'an-basin-like yi-ewer providesa further example of the handle-type under discus-sion. The handle-type may be claimed to be some-what rare but it is quite incorrect to assert that it is "never seen on any other vessel type."

7. This statement is not quite correct. One of the bronze vessels, at least, has the "imitation rivets" which Bagley refers to here. This is the Berlin vessel fully described by Sekino (op. cit., p. 11). It is also reproduced in T'ung-k'ao 2(1941), Pl. 486 — this is the best published photo and the "rivets" show up well; see also Sheng-kao(1940), Pl. 22. There appear to be two "rivets" and both are located longitudinally along the top of the extended ledge.

Since writing this note I have been able to study (May, 1980) the three Plates in Kummel's Chinesische Bronzen (Gesellschaft Fur OstasiatischeKunst 3. Jahresgabe. Berlin, 1928) two copies of which are in the Fogg Art Museum Library, Harvard University, Cambridge, Mass. Plate 6 is most valu-able for its detail on casting matters. What I took (and so, too, Sekino) to be "rivets... located longitudinally along the top of the extended ledge ... " are, in fact, so far as I can judge from the low-angle view, raised elements which form the "eyes" (one "rivet" as seen in the profile view) and the "eye-brows" (the second "rivet") of an animal-head with a nose-like ridge between (it would be very helpful to obtain a better view of the animal-head).

Although it is evident now that we cannot take these raised features to be "imitation rivets," the preceding clarification of the raised features does, nevertheless, seem to shed some light on what the bronze founders who produced the Berlin vessel may have judged the pairs of small dome-like protrusions (see examples in Figure 6) on the upper sections of handles in the pottery versions they copied to represent.

8. In Metallurgical Remains, the spout is includedamongst the several structural elements which I took to be indicative of ceramic copying of bronze orig-inals (cast, of course) in Early Shang Erh-li-t'ou contexts – see caption to Figure 3. In view of what – have written above (and the more thorough checking just recently undertaken as to the validity of the spout as evidence of inspiration from metal spouted vessels), I acknowledge the fact that insufficient thought was given to the point when Fig. 3 was being compiled. The six Erh-li-t'ou examples cited in the Figure include, as stated in the caption: "struc-tural elements presumably derived from metal proto-types manufactured at the same time but of which we presently possess no direct evidence." However, that there might be a problem as to which medium (clay or metal) first incorporated certain of the elements in question was not fully appreciated at the time.

9. In his paper, "The Art of the Chinese Metal-worker," distributed to the Symposium participants Tom Chase cites four long paragraphs — "part of a long stimulating letter from Robert Bagley (1977b)." In the present survey I refer to some of the tech-nical points included in these paragraphs. Other parts of the original letter, it would generally be assumed, would be for private rather than public reading. Such technical points as I refer to here (deriving from this source) were also read to the Symposium and made subject for discussion.

Regarding the term "crimping," correction is necessary. The method of joining-together visual-ized here (common enough in the modern world and we see it daily in cans, children's toys, biscuit tins, etc.) is that of the "folding" and "locking" of seams (note the frequent misuse of the term "seam"in casting parlance where some writers employ it in reference to mold joins). "Locking" is effected on a "pipe" or "stake" with a "seaming tool." If some such approach had been used in the hypothetical metal-worked li-ho to join the dome/spout unit andbody together, one would expect a form of "collar-ing" as used, for example, in wrought iron work. "Crimping" has nothing to do with the joining of two metal edges (see Oppi Untracht, Metal Techniquesfor Craftsmen and L.W.D. Ball and A.L. OdellPictorial Text Book of Engineering for further de-tails on all terms referred to here; the ConciseOxford Dictionary is very useful at times for suchterms — note "crimping" and its associations with textiles, the fluting of cartridge cases, etc.). To avoid confusion hereafter, the term "crimping" in Bagley's incorrect usage is maintained but presented between inverted commas. In cited passages, of course, it remains as is.

10. A thought which has developed in recent years during the series of casting experiments conducted

29

jointly with Wan Chia-pao and Ho Shih-k'un 何世坤 if at the Academia Sinica, Nan-kang 南港,Taipei (the first report, "The Casting of Inscrip-tions in Chinese Bronzes - with Particular Refer-ence to those with Rilievo Guide-lines" appears in the Soochow University Journal of Chinese ArtHistory 1 [Sept., 1976]:43-134) is that the natural tendency of the pottery clay to shrink as it dries, might have been utilized by the artisan in preference to the whittling down of the clay core surface. Degree of shrinkage may be controllable by varying the proportion of sand to clay, and possibly by other means. Appropriate experiments might lead to useful conclusions here.

11. Pottery facsimiles with such a curious handle structure do not appear either before, during, or after the period in which the bronze li-ho-kettleswere manufactured. Obviously, it would not appeal functionally to the ceramist and such an unwieldy appurtenance would be easily broken. Handles would be placed in such an awkward position in cast bronze vessels for only two reasons — (a) the hori-zontal mold division occurred there – thus allowing provision for an extension of the ledge to form the upper part of the handle, and (b) the handle was required to join on to this extension of the ledge — which was thickened slightly - for technicalreasons. The normal place to put the handle — as evinced in most other vessel-types — would be lower down directly onto the vessel wall (see Figure 6). The problem facing us is simply to determine why the thickening and extension of the ledge was made in the bronze li-ho. It is not a ceramic problem; andin the case of wrought metal receptacles, it should be appreciated that provision of a flap extending from the rim for the express purpose of attaching the upper terminal of the handle to it by means of rivets, would require very unusual approach to construction. So unusual that one may justifiably doubt that a metal-working artisan would attempt so problematic and potentially ineffectual a method of join. The very fact that a search throughout an appreciably large corpus of relevant publications in the European and Middle East Bronze Age areas has been singularly unsuccessful in finding a single example of handle joins of this peculiar kind, leads one to suppose that no such method of metallurgical join was ever utilized in any metallurgical culture. Then in the Chinese ceramic scene, other than the Andersson vessel, those with "rivets" generally lack the "flap" element.

12. I have left the present paragraph exactly as written in the early draft. Later examination of the P'an-lung-ch'eng vessel at the Exhibition (through the display-case) and a re-examination of the two Brundage vessels (made directly on the ves-sels) has demonstrated that the tentative observa-tion offered earlier above now has some real foun-dation. The paragraph preceding the present one, however, has been completely re-written — these recent examinations, particularly of the two Brundage vessels, showed that appreciable revision would be required in places.

13. To state, as Bagley has done, that "(The lower end of the handle, which need not be so firmly joined to the vessel, would typically have been soldered on or secured by an adhesive. The spout

would have been attached in the same way)" (see extract from the Bronze Age Catalogue in Post-script) is rather imaginative to say the least.

13a. See Postscript for more detailed reference to the relevant content of this sumptuously produced Catalogue covering the Exhibition of the same title currently on display in the USA. Very use-ful essays by Ma Ch'eng-yuan 馬承源 ChangKwang-chih 張光直, Robert L. Thorp, and others,and detailed notes on the bronzes and other items on exhibit by Robert W. Bagley, Jenny So, and Maxwell K. Hearn appear. A great deal of work has gone into the preparation of this publication and it contains much valuable information as well as numerous fine illustrations in color. Leaving aside the matters raised in the Postscript, I would be inclined to offer only one major critical observation in respect to the Bronze Age Catalogueas a whole: the unfortunate decision to employ the Quixotic p'in-yin 拼音 romanization which makesthe work both difficult and distracting to read. Those who were responsible for this decision should, I feel, have given greater thought to the unhappy effect the p'in-yin spellings have upon the generalpublic (see my assessment of the p'in-yin problemin Barnard and Cheung Kwong-yue 張光裕, Studiesin Chinese Archaeology, 1980 — Reports on Visits to Mainland China, Taiwan, and the USA; Participa-tion in Conferences in these Countries; and some Notes and Impressions (in press, to appear circa May, 1981).

14. In the preparation of Figure 8 I used as an example one of the P'an-lung-ch'eng ting-cauldrons,checking such photographic data that was then avail-able in several Chinese Exhibition Catalogues. Usefully, the "Great Bronze Age of China Exhibition" at the Metropolitan. Museum, had the same item (No. 4) on display and also the fang-ting, No. 11.Though the depth of focus of the former vessel photograph (Bronze Age Catalogue, p. 84) is notwell achieved — and one can sympathize with the photographer's problem here — the rear view of one handle and the immediate rim surface alongside it demonstrates clearly enough the technical points made above. In his descriptive notes, however, Bagley has again pushed forward his wrought metal-working theories: "The outer half of the wide everted rim steps up slightly, as though the entire rim had been hammered out and then folded inward to give a smooth edge. This imitation in cast bronze of a feature rising from wrought-metal tech-nology has already been noted in connection with the jue [chüeh] no. 1." Earlier in the Bronze AgeCatalogue (pp. 70-73) a more detailed descriptionof mold preparation and casting method appears in relation to this vessel which is almost correct, so far as it goes, but marred by an inaccurate "sche-matic drawing" of the mold-assembly. The drawing lacks funnel-shaped pouring inlets such as would be required for pointed members of this kind — (see Figures 8, a and b); the minute entries provided for each leg would tax the efforts of a founder to run in the molten bronze! Further, the drawing lacks mortice/tenon style locking devices on the mold section edges.

15. "Edge thickening" is also known as "upsetting" or "swaging" and involves the striking of "the·

30

vessel edge with a light-weight collet hammer with light, over-lapping blows delivered directly against the edge at a right angle to the face of the hammer"(Oppi Untracht, p. 248). As to the antiquity in the West of the three methods of thickening, or rein-forcing, the bowl edges noted above, I believe my statement to be correct in view of photographs of relevant items handy at the moment of writing.

The phrase "outwards and under" is emphasized above as this appears to be an almost universal rule in the West, and no doubt one dictated by technical considerations, i.e., the direction of forming the bowl is towards the mouth edge and flaring outwards, so the natural direction of folding along the edge (either a simple fold, or one to fold over reinforce wire) must be outwards and under.

There may be some ambiguity attending Bagley's identification of "folded rims." I am not certain whether the thickened edges of the bronze li-ho rimsare also covered by this term. If so, it would be necessary to point out that this feature in the bronze li-ho-kettle has nothing to do with wroughtmetal "folded" or "reinforced" rims, and it is not related to the mold/core parting-line which I have just discussed. It is either derived from a spec- ifically ceramic aspect of rim construction — a thickening of the material of the rim in order to maintain the stability (and strength) of the mouth or it may be simply a decorative feature; in all probability it is a combination of both. In the case of the two Brundage li-ho, it may be noted,there is little or no thickening of the metal edge; the feature is effected simply by an intaglio line paralleling the edge.

16. Noticeably, the terms "cold working" and "cold hammering" do not appear in his published paper, while "annealing" appears once only and this in a cited passage in a footnote (p. 198, n. 46).

17. There is neither space nor time to consider here other technical matters which Bagley has touched upon in his article, e.g., cire-perdue,the origin of the flange, the significance of posts on such vessels as chüeh and chia. The implicationis not that his observations are largely at fault but rather that some interesting points arise from them, in particular, the origin of flanges which I will take up later in another paper.

18. How the ancient founder could cast so thin-walled a vessel was the main theme and we could do little other than observe the fact that they were obviously able to do so. Earlier, before Cheung Kwong-yue 張光裕 and I visited the Mainland inMarch, 1980, Kao Ch'ü-hsün 高去尋, Director ofthe Institute of History and Philology, suggested I should keep an eye open for any evidence of metalworking in pre-Han bronzes. Later, after the International Conference on Sinology (August, 1980) organized by Academia Sinica at Nan-kang 南港,both Tom Chase and I had an interesting session with Kao, Shih Chang-ju 石璋如, Wan Chia-pao, and HoShih-k'un during the course of which fragments of a very thin-walled bronze yi-ewer 匜 with finely in-cised decor on the inside surfaces which came from

the early digs at Hui-hsien 輝縣 were shown to usfor comment on method of manufacture. Could this be an example of metal-working? Similar thin-walled vessels are known, e.g., a fragmentary yi-ewer inthe Seattle Museum (Ch.6.86) and the well-known fragmentary Ch'ien-basin 鑑 (1:73) similarly cov-ered with incised decor in the inside wall, in Hui-hsien fa-chüeh pao-kao 輝縣發掘報告 (1956),p. 116. The fragility of the vessels — the last one with contemporary brazed (?) repair traces (p. 115) — suggests immediately cast manufacture rather than worked metal. Other details such as the nature of the thickened rim and the use of cast protruberances affixed "rivet"-fashion to the vessel walls seem to indicate direct casting in molds. Laboratory examination is, of course, required to confirm these observations. It is from about this period (Ch'un-ch'iu/Chan-kuo) that the introduction of metal-working techniques may be expected to appear in China.

19. Over and above technical considerations such as those with which we are concerned here, attention would need to be drawn to a number of mishaps attend-ing the inscription translations. However, these are not really relevant to the theme of the present survey and may receive attention elsewhere. So far as art-historical matters are concerned I have little expertise. But in view of the rather criti-cal nature of my observations, it would be unfair to omit reference to the credit side of Bagley's contribution — and a very large contribution it is — to the Catalogue. Obviously, a great deal of effort has gone into the venture and a great deal of thought as to the significance of the materials on exhibit in their art-historical perspective.