The dating of the Chu graves of Changsha

Acta Orientalia (Copenhagen), 1987, 48: 111-156.

Donald B. Wagner

 

• 1. Excavations of ancient graves in Changsha
• 2. The dating of the graves
• 3. Grave forms
• 4. Ceramic artefacts
• 5. The site of the Chu capital in Jiangling, Hubei
• 6. Iron in Jiangling and Changsha
• 7. Iron in the state of Wu
• 8. References and abbreviations
• Notes

---

Until very recently nearly everything that was known archaeologically of the ancient Chinese state of Chu 楚 came from excavations in the vicinity of Changsha, Hunan. Some 2000 Chu graves have been excavated here. A chronology for these graves which is widely but not universally accepted places about 100 of them in the Spring and Autumn period: this dating implies that some twenty iron artefacts found in the graves are to be dated to ca. 500 B.C. or before. These would then be among the earliest iron artefacts (other than meteoritic iron) yet found in China.

It happens that the question of the first use of iron in China is important to me, and I have therefore studied the Changsha material rather carefully. This study has forced me to the conclusion that the usual chronology of the Chu graves is faulty. and that currently available material does not permit a better dating for these graves than "pre-Han''.

The purpose of the present article is to argue this negative conclusion in detail. The final sections then reconsider briefly the evidence concerning the cultural context of the first use of iron in China.

1. Excavations of ancient graves in Changsha

Antiquities from the ancient graves of Changsha have since the 1930s been bought by collectors at high prices, and grave-robbery has often been a lucrative sideline for the people of the area. Changsha pottery''. for example, is a standard category in the art trade; it can be recognised by characteristic stains from the heavy red clay soil of the region. As late as 1951 antiquities from Changsha were still being legally exported to Hong Kong (tregear 1961; Cheng Te-k'un 1963: 163; Eliseef 1976: 1; Yu Weichao 1984: 11-16).

Xia Nai (1952: 493) estimated that over 1000 Changsha graves had been destroyed before the founding of the People's Republic of China in 1949. From that time, with the coming of peace and renewed construction, ancient graves were found in increasing numbers in the course of building construction, earth-levelling, and clay-quarrying for brick manufacture. The scientific excavation of these graves became one of the first priorities of the reestablished Institute of Archaeology of the Academia Sinica. A work team was sent to Changsha. and its excavation report (Changsha 1957) is still the most precise and detailed account available of the Changsha graves.[1]

Salvage excavations were carried on from October 18,1951 to February 7,1952; 162 ancient graves were excavated, and these were dated as follows:

Warring States 73
Early Western Han 27
Later Western Han 38
Eastern Han 7
Tang - Song 15
undateable 2

In the seriation the boundary between Warring States and Early Western Han is marked by the appearance in many graves of clay imitations of Han coins. Later Chen Zhi (1961) pointed out that these are imitations of coins issued under the emperor Wen-di (r. 179-157 B.C.); thus some of the graves dated Warring States may actually be from the early decades of the Han.

The Academia Sinica excavations ended in February 1952: in May 1952 the Changsha Suburbs Work Team for the Salvage of Ancient Graves was established to continue the work using local personnel. Its work was taken over in January 1953 by a body responsible for salvage archaeology in all of Hunan Province. As in so many salvage archaeology projects in other parts of the world, the team worked under difficult conditions. It was inexperienced and undermanned, and met considerable resistance from construction workers who resented the delays which archaeological work caused them. one of the participants wrote:

Construction work in the Changsha suburbs is being carried out over a large area. Earthwork operations are the responsibility of four offices. for the eastern, southern, western, and northern suburbs. Each office leads several companies, and each company consists of about 500 workers. The earthwork proceeds at astonishing speed, and a small mountain can be levelled in a few days. Every day many ancient graves are exposed. and some are completely destroyed. The agency responsible for the archaeological work is the Hunan Provincial Work team for the Clearing of Ancient Graves, which has only about ten workers. Every day they race about among the four suburbs, clearing the ancient graves which have been exposed. . . . (Cheng Xinren 1953: 43-44)

Besides the race with construction work there was also competition from another quarter: in 1952 alone over 200 ancient graves were destroyed hv an organised group of treasure hunters (Wu Mingsheng 1953: 46).

---

There seem to have been numerous large rich Chu tombs. covered by mounds. in the Changsha area, but nearly all of these have been robbed and are therefore without much archaeological value (Li Zhengguang & Peng Qingye 1957: 33). Very few excavations of large Chu tombs in Changsha have been published. For convenience I use the term tomb here to refer to these, and grave for the smaller and poorer burials which are the subject of this article. These graves are numerous, and have been less subject to robbery: therefore they represent the most important material for Chu archaeology in Changsha.

By 1979 over 2000 Chu graves had been excavated in Changsha (Sanshinian 1979: 312-313). At present we have, besides the Academia Sinica report, a detailed excavation report on 61 Chu graves excavated north of Changsha (Li Zhengguang & Peng Qingye 1957) and a rather cursory summary of material from 209 Chu graves excavated from 1952 to 1956 in the eastern and southern suburbs (Wen Daoyi 1959). Descriptions of some individual graves, of special interest for one reason or another, have also been published. Two forthcoming books, when they become available. should provide a wealth of material for systematic study; one of these is a report on 1056 graves.[2]

The Chu graves of Changsha are vertical rectangular pits with dimensions as small as 2.00 x 0.56 m and as large as 4.80 x 3.75 m, depth from 0.70 to 9.80 m (Changsha 1957, tables 1-3). The fill is very solidly tamped earth. Pedological conditions in the area are such that organic matter in the graves is preserved only in exceptional cases. Usually there is no trace of human remains or coffins, but it is believed that most graves originally contained wooden coffins (guan 棺)and outer coffins (guo 椁). Some have a ledge or "second-level platform" (ercengtai 二層臺), and this may be a substitute for an outer coffin. In some graves a "niche'' (bikan 壁龛) is dug out of the wall of the pit for the placement of grave-goods. (For explanations of these and other common features of ancient Chinese graves see K. C. Chang 1977: 357-366.)

The preserved grave-goods include a variety of bronze and ceramic artefacts and a few artefacts of iron, wood, and lacquer. It seems likely that wood and lacquer grave-goods actually were fairly common. but have been preserved only in exceptional cases.

The ceramic artefacts are most often of very poor quality, made of gritty unwashed clay and fired at a low temperature. The Academia Sinica report states: "In many cases, when they were first excavated they could be identified as to form, but afterwards, because the body was as soft as mud, they could not be reconstructed" (Changsha 1957: 3, 28). Presumably these were cheap articles made especially for use as grave-goods. Within each type of ceramic artefact the Academia Sinica report defines a number of styles, e.g. guan-jar styles l-lV, ding-cauldron styles l-VI, etc.; however the detailed table of graves often lists types without style numbers: in these cases it seems that the artefacts had been identified in the field but disintegrated before the detailed classification work began. I have not seen this problem mentioned in reports on the later excavations, but there is every reason to believe that it has continued to be a problem which severely hampers efforts to develop a local chronology based on ceramic artefacts.

2. The dating of the graves

How are these graves to be dated? An article by Gao Zhixi (1979) on the Chu graves of Hunan indicates that three very important dating techniques are not applicable here.

For the relative chronology of the Chu graves of Hunan there is no hard data from stratigraphy or disturbance of one grave by another. Apparently there are no clear strata in the Hunan soil for the period under discussion; furthermore many of the graves have been found in earth-moving operations which have destroyed the mouths of the graves and therefore whatever stratigraphic relations there may have been.

There are no early bronze inscriptions to give absolute dates.

The contents of the graves are generally too poorly preserved to provide the amount of organic material necessary for radiocarbon dating. Furthermore radiocarbon dates normally have such broad uncertainty limits that they could not in any case be expected to be of any assistance in our present inquiry. Exactly one radiocarbon date has been reported for a Chu grave in Hunan. and it happens to be the first ever reported in China. no. ZK-I (KG 1972.1: 55; Tan shisi 1983: 98; see also Barnard 1975: 10-11; Rudolph 1973): This gives a date of 2330+90 b.p. for a wooden figurine from grave no. 406 of the Academia Sinica excavations (Changsha 1957: 25-27). Using the dendrochronological calibration of Klein et al. (1982) this gives, at the 95 per cent confidence level, a range of possible dates between 765 and 175 B.C., which in the present inquiry is of no use at all.[3]

---

Since there is no doubt that these are Chu graves it would be very useful in this discussion to know, from archaeological or written sources, when Chu settlement reached as far south as Changsha, 250 km south of the Chu capital in modern Jiangling, Hubei. Shang remains have been found in Changsha as well as in various other areas in Hunan, but these clearly reflect Shang culture (with some local peculiarities). and are not considered to be related to Chu culture (Zhou Shirong 1979; Gao Zhixi 1981; see also Kane 1975).

The earliest independently-dateable evidence for Chu settlement in Changsha is a very large and rich Chu tomb at Liuchengqiao 浏城桥. In it were found 262 artefacts, all extremely well preserved, and these have provided the basis for what seems to be a fairly reliable dating. The excavation report suggests that it is to be dated to the fifth century B.C. (KGXB 1972.1: 71), and Gao Zhixi (1982) argues that the date is more precisely the early fifth century B.C. A study by Chen Zhenyu (1981: 327) suggests an early Warring States date, i.e. in the late fifth century B.C. The latter dating is accepted by the Jiangling excavators (Jiangling 1984: 135) without comment.

The Liuchengqiao tomb thus provides good evidence that Chu settlement had reached Changsha by the end of the fifth century B.C. Can written sources take this farther back? The evidence has been reviewed by a number of scholars (e.g. Gu Tiefu 1982; He Hao & Yin Chonghao 1981), and the result is disappointing: none of the early sources so much as mentions Chu expansion south of the Changjiang. The earliest sources which could be found are three geographical works of the Tang period. Each of the three indicates that at the end of the Spring and Autumn period the territory of Chu included all of modern Hunan. I have two objections to the use of these sources: (1) it is not clear what major sources the Tang authors could have had that we do not have now; (2) the three give this information in almost exactly the same words. so they are clearly not independent sources. We may well wonder whether anyone in the Spring and Autumn period, even the rulers of Chu themselves had any very clear idea of how far south Chu rule extended. or any precise knowledge of the geography of the region south of the Changjiang.

The most useful statement in the written sources concerns the minor state of Luo 罗, which is mentioned briefly in the Zuo zhuan and the Guo yu. [4] The Zuo zhuan describes an attack by Chu in 698 B.C. which was repulsed by Luo with the aid of the Lurong 卢戎 barbarians. The Han shu (1962, 28a: 1566) states that the Han prefecture of Zhijiang 枝江 (modern Zhijiang County, Hubei) is the location of ancient Luo. The commentator Ying Shao 應劭 (d. ca. A.D. 204) states that King Wen of Chu (r. 689-677 B.C.) moved Luo to the Han prefecture of Luo, in modern Miluo County, Hunan (Han shu 1962, 28b: 1639). Du Yu 杜預 (A.D. 221-284), commenting on the Zuo zhuan passage, states that Luo was in Yicheng 宜城, modern Yicheng County, Hubei. in 699 B.C. and was later moved to Zhijiang (SSJZS 1756).

The important piece of information here is the statement by Ying Shao that Luo was "moved'' by Chu in the early seventh century B.C. to Miluo, about half-way between Changsha and the Chu capital in Jiangling, Hubei (see Section 5 below). This appears to be the earliest source which specifically mentions Chu activity south of the Changjiang in the Spring and Autumn period. What actual events lie behind the story can only he guessed at: but an ancient city-site has been found and cursorily excavated in Miluo County, and has been identified by the excavators with the capital of Luo (Zhou Shirong 1958: K. C.Chang 1977: 427).

Material from this excavation is scant. but seems to indicate that the material culture of Luo was not very different from that of Chu. Our interest here is in the expansion of a culture rather than a political entity, and it is likely that Chu political expansion followed rather than preceded settlement by "refugees" or "homesteaders" such as these. We have no way of knowing whether Ying Shao had a reliable source for his statement, but if we choose to believe him this was in the early seventh century B.C. There is then nothing very unlikely in the assumption that Chu culture reached Changsha about the same time, or even before, but proof seems to be entirely lacking.

Since there is no help to be had from stratigraphy, disturbance between graves, radiocarbon dating, or historical sources. the only possible way of dating the small poorly-preserved Chu graves of Changsha is to develop a local relative chronology based on grave forms and artefact styles, then attempt to find a few absolute dates within this chronology by comparing artefacts with well-dated artefacts from excavations elsewhere.

Gu Tiefu (1954) was the first to attempt such a dating. By 1954 several hundred Chu graves had been excavated in Changsha, and by far the most common complement of ceramic grave-goods was found to be: ding-cauldron - dui-bowl - hu-vase. Figure 1 shows some examples of the ding-cauldrons. The ceramic vessels of this artefact complement are all clearly imitations of bronze vessels.

However a very few (I0+) graves had been excavated which instead of these types contain one or another combination which includes a li-cauldron. The commonest of these combinations was: li-cauldron 鬲 - bo-bowl 缽 - small flat-bottomed hu-vase 壺. Figures 2-3 show some examples of li-cauldrons and bo-bowls. Artefacts in these combinations are generally of poorer quality paste than the former: further, the graves in which they are found tend to be much smaller, and only rarely contain bronze artefacts. Gu Tiefu therefore assumes that graves which contain li-cauldrons are earlier than those which contain ding-cauldrons.

Gu Tiefu then attempts to provide an absolute date for the graves which contain li-cauldrons. He notes first that the well-dated li-cauldrons found in north China are so different from those found in Changsha that no comparison is possible. The reader can easily verify this by comparing Figure 2 with K. C. Chang's illustrations (1977: 307, 316) of Zhou li-cauldrons. Yu Weichao (1980: 10-11) states categorically that this style of li-cauldron cannot have developed from Shang or Zhou styles.

Gu Tiefu sees the key to the dating of the early graves in grave no. 52.826, which was excavated in 1952. It contains a li-cauldron, It bo-bowl. and a fragment of what might be a hu-vase. It also contains a bronze sword, a bronze mirror, and an iron scraper-blade (see Figure 4.1). The mirror is of what seems to be a very primitive type: it is undecorated. is very thin (less than 2 mm). and has a very small eyelet (ca. 5 mm). For these reasons Gu Tiefu considers this to be the oldest type of mirror.

In 1954 it was still widely believed that the mirror was a Warring States innovation (cf. K. C. Chang 1977: 370). Gu Tiefu therefore considers that this oldest type of mirror cannot be later than the beginning of the Warring States period. Thus the li-bo-hu graves cannot be later than this time, and iron was in use in Changsha at this time.

The logic of this argument is obviously less than impeccable: the Academia Sinica excavation report quotes Gu Tiefu's conclusion and dismisses it with the words, "This remains to be proved" (Changsha 1957: 36).

Later a special study of ancient undecorated mirrors was made by Wang Zhongshu (1963). He studied 23 mirrors for which some sort of date was available and divided them into four styles: Gu Tiefu's mirror, being completely plain, is placed in style 1, which includes eight mirrors in all. Six of these are dated Warring States and one is dated Western Han. Thus plain undecorated mirrors were in use as late as the Western Han period, and Gu Tiefu's argument is demolished. Finally it should be noted that material from more recent excavations has put the origin of the mirror in China all the way back to the Shang period (K. C. Chang 1977: 370-371).

Gu Tiefu's article established the two basic assumptions which have been applied in later attempts at a relative chronology of the Chu graves of Changsha: small graves are earlier than large graves, and graves containing li-cauldrons are earlier than those containing ding-cauldrons. Chronologies by workers in Changsha have almost always preserved the early date for the use of iron suggested by Gu Tiefu, while others have been more sceptical.

Wen Daoyi (1959: 56), in a survey of 209 pre-Han graves in Changsha. places long narrow graves containing cauldrons in the earlier of two periods, and suggests that "the earliest limit of this period cannot be later than the beginning of the Warring States, and may possibly extend to the end of the Spring and Autumn period.'' His argument for this dating generally follows that of Gu Tiefu: the li-cauldrons cannot be directly compared with northern li-cauldrons, because their style is so different; but some of these early graves contain small plain mirrors. "Therefore our placement of Changsha graves which contain ceramic li-cauldrons in the early Warring States or the end of the Spring and Autumn period is certainly not entirely without basis.''

Gao Zhixi's general description (1960: 33) of the ancient graves of Hunan (from the earliest times to the Ming) states that about 100 Chu graves are small and generally have niches. The most representative ceramic artefacts in these graves are li-cauldrons, lei-jars 壘, and dou-bowls. He states that these graves are to be dated to the Spring and Autumn period, at the latest to the beginning of the Warring States. He does not in this article give the reason for this judgement, but in a review (1962: 46-47) of the Academia Sinica excavation report he notes that various Neolithic,[5] Shang, and Western Zhou sites in Hunan have yielded li-cauldrons and dou-bowls. These are in fact irrelevant, for it is generally agreed that these sites are not related to Chu culture. His best argument is that the li-cauldrons and dou-bowls found at the site of the city of Luo (mentioned above), which is dated to the Spring and Autumn period, are very similar to those found in this "early" group of graves. The same argument is also given in a textbook of archaeology published in Changsha (Changshi 1959: 54-55). Turning to the excavation report on the Luo site (Zhou Shirong 1958), it turns out that no complete ceramic artefacts were found, only rather small sherds; that the site is dated by comparison of these with the Changsha artefacts; and that the only date the excavators actually give is Eastern Zhou. Assuming that this really is the site of Luo, and that the city was established in the seventh century B.C., there is still no evidence as to when the site was abandoned, except for a Han kiln which disturbs it; thus a useful date for the sherds cannot be arrived at independently.

The following Sections discuss the possibilities for relative and absolute dating presented by grave forms and ceramic styles respectively.

3. Grave forms

In the Academia Sinica excavation report the 73 graves considered to be pre-Han are divided into groups as follows:

1. Narrow rectangular pits, 39.
1.1. Seven of these which are especially narrow. with the width less than one-third of the length.
1.2. Thirty-two others.
2. Rectangular pits with access ramps, 21.
3. Wide rectangular pits, 13.

The report states that this grouping by grave form is not in general relevant to the relative chronology, but suggests that group 1.1 graves. which contain rather different types of ceramic artefacts, may be "slightly earlier" than those of the other groups (Changsha 1957: 34-35, 36, 161-162). An iron spade-cap (Figure 4.3) was found in one of these. and thus iron would appear to have been used in Changsha in this earliest period.

Statistical methods are in order here, and when detailed data on the 2000-odd Chu graves of Changsha have been properly published we will have an excellent sample for statistical analysis. Only two homogeneous and detailed bodies of data are now available: for the 73 pre-Han graves of the Academia Sinica excavations, and for the 61 pre-Han graves reported by Li Zhengguang and Peng Qingye (1957). Detailed data is available for perhaps 50 more in reports of individual graves. but each of these has been reported for some specific reason, and therefore they do not constitute an unbiased sample for statistical purposes .

The grave-goods and dimensions of the graves in group 1.1 are tabulated in table 1 . For comparison the grave-goods and dimensions of graves in the other three groups are summarised in the same table. It should be explained that an entry like "2.18 + 0.14" in the next-to-last column indicates that the mean (average) of the lengths of graves in group 1.1 is 2.18 m and that the associated standard deviation is 0.14 m. The standard deviation is a statistical measure of the "spread" of the data away from the mean; it does not indicate the absolute minimum and maximum of the lengths.

table 1. table of grave-goods and dimensions of 73 Chu graves excavated in Changsha (derived from Changsha 1957: 169-170). For the definitions of the groups see Section 5. Graves in group 1.1 are described individually; for each of the other groups the total number of artefacts of each type is given, together with the mean dimensions and the associated standard deviations. The last column gives the ratio of bottom length to bottom width. All grave-goods found in group 1.1 are listed: many artefact types found in the other groups are ignored.

group	grave no.	ding -cauldron	dui-bowl	hu-vase	wan-bowl	bo-bowl	guan-jar	bei-cup	dou-bowl	li-cauldron	bronze sword	bronze ding-cauldron	bottom dimensions, metres	L/W
1.1	111			1	1	1							2.34 x 0.62	3.77
	117												2.30 x 0.70	3.29
	301					1					1	1	2.30 x 0.66	3.48
	305			1		1		1					2.20 x 0.60	3.67
	308						1		3				2.00 x 0.56	3.57
	314					1	1			1			2.10 x 0.68	3.09
	316					1	1			1	1		2.00 x 0.60	3.33
	total/mean	0	0	2	1	5	4	1	3	1	2	1	2.18+/-0.14 x 0.63+/-0.05	3.46 +/-0.24
1.2	total/mean 32 graves	35	29	30	2	0	2	0	8	0	6	1	2.59+/-0.27 x 1.14+/-0.26	2.37 +/-0.48
2	total/mean 21 graves	14	6	10	0	0	1	0	4	1	5	1	3.31+/-0.56 x 2.22+/-0.60	1.53 +/-0.19
3	total/mean 13 graves	22	17	24	0	0	10	3	3	0	5	2	3.03+/-0.33 x 2.00+/-0.45	1.56 +/-0.23

The last column of the table gives the ratio of length to width (L/W) at the grave bottom. This ratio is probably the most useful of the statistics given, since it represents the visible shape of the grave. There would likely have been more tolerance in the actual length and width as long as the grave "looked right" according to the judgement of the persons involved. Since the walls of some graves taper significantly it would have been preferable to use here the dimensions of the mouth of the grave rather than the bottom, but in 21 of the 73 graves the mouth was destroyed by earth-moving operations before the archaeologists arrived on the scene.

The histograms in Figure 5 show the distribution of L/W for groups 1. I and 1.2 and for groups 2 and 3 taken together. Most readers will be content to note here that each histogram shows a distinct peak and what appears to be a "natural" tapering off on either side of the peak: and that the groups show distinct differences. Those who are familiar with statistical methods will see that each of the distributions is fairly close to a normal distribution ("bell-shaped curve''), and that the differences between groups 1.1 and 1.2, and between group 1.2 and the remaining two, appear to be statistically significant.

The most likely explanation for these histograms is that the four groups represent three distinct ideal grave-shapes which the persons concerned attempted to approximate.

---

Five of the graves in group 1.1 contain bo-bowls, whereas none was found in any of the others. The guan-jar is common in this group and rare in the others. One li-cauldron was found in this group, and only one other was found. In the other groups ding-cauldrons, dui-bowls, and hu-vases are common; in group 1.1 only two hu-vases were found, and no ding-cauldrons or dui-bowls at all.

Group 1.1 is thus considerably different from the other groups by two independent criteria, grave form and ceramic artefact types. If the graves of group 1.1 are to be placed in any sort of chronological relationship to the others some development must be shown among the groups. Here we see, in two important parameters, not a development but three apparently unrelated complexes of burial practices: the shapes of the graves are significantly different, and the grave-goods are of different types. Therefore no conclusion can be drawn as to the relative ages of the groups.

---

The only other body of data now available on Changsha Chu graves which is amenable to statistical treatment is the report of Li Zhengguang and Peng Qingye (1957) on 61 pre-Han graves cleared in salvage excavations in the Shahuqiao 沙湖橋 district. The data for 59 of these (for two graves the dimensions are not given) can be divided into groups in the same way as above:

Plotting the data in histograms (not given here) shows the same sort of approximately normal distributions as in Figure 5. An obvious difference is that group 2+3 graves are much less common (15% vs. 47%) while group 1.2 graves are more common (76% vs. 44%). Another difference is that four of the five graves in group 1.1 (A21, A38, A59, and E5) contain the usual complement of grave-goods: ding-cauldron, dui-bowl, and hu-vase. The remaining grave in this group (A35) contains a spearhead and a sword, both of bronze, and no ceramic artefacts at all.

These minor differences between the Academia Sinica data and the Shahuqiao data are probably to be attributed to regional differences in burial practices within the Changsha area. Shahuqiao is about 2.5 km north of the city, whereas in the Academia Sinica excavations 48 of the 73 pre-Han graves were in the eastern suburbs, and 25 were at Wujialing 伍家陵, in the northern suburbs. Of these 25, 16 graves (64%) were in group 1.2, 2 were in group 2, and 7 were in group 3 (36% in groups 2+3): We know virtually nothing of social life in ancient Changsha, but it seems reasonable to presume that the population was thinly scattered over a large area, and that traditions concerning burial of the dead varied to some extent from place to place within this area.

4. Ceramic artefacts

In more recent work on Chu grave chronology grave forms have not been mentioned - perhaps because, as examples multiplied, it became apparent that the relation between grave forms and artefact types was more complex than had been observed earlier. At present attention is directed almost solely to ceramic artefacts. The most ambitious and extensively argued attempt so far at a chronology of Chu ceramic styles is that of Gao Zhixi (1979). Here he takes all the Chu graves found in Hunan together. and gives the chronology translated in table 2. He is not very specific on how this table is to be understood. The list of ceramic types under each period appears to be an ideal complement to be expected in a grave of the given period, of which usually only a subset appears. For example only one grave contains the full complement listed for period 1, and none contains the full complement listed for period 2 (see his table, 1979: 242). In operational terms it seems that the following rules are to be applied:

1. If the grave contains a li-cauldron, it is to be placed in period 1 or 2 according to the style.

2. If it contains a lei-jar it is to be placed in period 3.

3. If it contains any one of the ceramic artefact types which normally accompany a li-cauldron or a lei-jar, it is to be placed broadly in periods 1-3.

4. If it contains a ding-cauldron and a dui-bowl, it is to be placed in period 4 or 5 according to the style of the dui-bowl.

5. If it contains a ding-cauldron but not a dui-bowl, and in addition one of the ceramic artefacts listed under period 6, it is to be placed in that period.

---

Table 2. Gao Zhixi's proposed chronology of ceramic grave-goods excavated in Hunan Province (1979: 238).

Period 1, Middle Spring and Autumn period [7th century B.C.] li-cauldron 鬲, style I bo-bowl 缽 (also referred to as a pen-basin 盆 guan-jar 罐 hu-vase 壺 dou-bowl 豆

Period 2, Late Spring and Autumn period [6th century B.C.] li-cauldron, styles II and III bo-bowl guan-jar hu-vase

Period 3, Spring and Autumn - Warring States transition [beginning of 5th century B.C.] lei-jar 壘 (also referred to as a bu-jar 瓿) bo-bowl dou-bowl (The above three periods also include some graves containing hu-vases, bo-bowls, dou-bowls, and guan-jars.)

Period 4, Early to early-middle Warring States period [ca. 470-350 B.C.] ding-cauldron 鼎 dui-bowl 敦 (oval or spheroidal) hu-vase possibly also: dou-bowl [replacing dui-bowl: see 1979: 246]

Period 5, Late-middle Warring States period [ca. 350-300 B.C.] ding-cauldron dui-bowl hu-vase possible also: fang-vase 鈁 pan-basin 盤 yi-pitcher 匜 shao-ladle 勺

Period 6, Late Warring States period [3rd century B.C.] ding-cauldron he-bowl 盒 hu-vase fang-vase possibly also: xunlu-incense burner 薰爐 pan-basin yi-pitcher shao-ladle

---

Gao Zhixi gives photographs of examples of li-cauldrons of his styles I and II, but these would not reproduce well here. He describes the three li-cauldron styles as follows (1979: 242):

1. The bottom is convex inward. The crotch is shallow, but still apparent. The feet are conical and relatively high. Except for the lip the whole body is cord-patterned.

Il. The bottom is nearly flat, and the crotch is very shallow. The legs are cylindrical. The neck and belly are cord-patterned.

III. The bottom is round and the size is relatively large. The legs are solid and have an elliptical cross-section. There is very little cord-patterning.

In Figure 2 the rightmost li-cauldron is almost identical to Gao Zhixi's example of style I. The leftmost is presumably also to be classified as style I, though it does not seem in the drawing to have cord-patterning. The one in the middle would seem to be somewhere between styles I and II. I have not been able to find an illustration of a li-cauldron which fits the description of style III.

In discussing this chronology we may start with Gao Zhixi's argument (1979: 243) that the Chu graves of Hunan do not go farther back than the middle Spring and Autumn period. A Chu grave in Xiangxiang (70 km southwest of Changsha), which he places in period 1, disturbs a pre-Chu grave which he places in the early Spring and Autumn period. Therefore the graves of period 1 cannot be earlier than the middle Spring and Autumn. This is the entire argument.

We have here in its purest form an assumption which pervades nearly all work on Chu chronology in Changsha: in each period only one type of grave occurs. In this case, one grave of a certain type is disturbed by one grave of another type; therefore all graves of the latter type are later than all graves of the former type. No account is taken of the possibility that there may have been a long period of overlap in which both types of grave occurred. Even if this assumption were allowed, it would still be unreasonable to assume that the two grave-types neatly fall into the arbitrarily-defined periods "early" and "middle" Spring and Autumn. We have no way of knowing how many years passed between the two burials. As was discussed in Section 2 above, there is not much evidence one way or the other concerning when Chu culture reached Changsha, and this one disturbance of a pre-Chu grave by a Chu grave is not evidence that this did not occur as early as the eighth century B.C.

Considering now the relative chronology of table 2, we see that it is based on the assumption that the entire population, over a period of five centuries, all buried their dead in the same way and changed their burial practices at the same times; and that each change was completed within a very short time. Especially odd here is period 3. This period is characterised by the ceramic lei-jar, which occurs in no other period and is quite different from the other vessel types listed. According to table 2 it was placed in graves for a period of only a few decades, and in this period tripod cauldrons (li or ding) did not occur. All this is simply not in keeping with what we know of real human cultures.

It is also an error to attempt to place all of the Chu graves of Hunan in a single seriation. As a standard textbook on archaeological method puts it, "Generally, the more limited the source area of the artifacts - such as a small valley or a single site, as opposed to an area such as 'southern France' - the more reliably the seriation relates to time changes" (Sharer & Ashmore 1979: 361). This stricture would apply especially to the ceramic artefacts in question here, since they undoubtedly were locally produced for local use. bronze ritual vessel styles, on the other hand, seem in Chinese archaeology to provide more reliable material for seriation on a larger geographic scale.

Archaeologists usually postulate that any particular attribute follows a career in which its frequency starts small, increases to some maximum, then decreases until it finally goes completely out of use again. The time-span of the whole process from initial appearance to maximum popularity to disappearance again, as well as the maximum frequency attained by the attribute in question, can be expected to vary greatly from attribute to attribute. This postulate has been shown to be valid in contexts where the relative chronology is given by some independent criterion such as stratigraphy (see e.g. Sharer & Ashmore 1979: 360-363). Thus we should expect considerable overlap in the uses of different types of ceramic artefacts in Chu burials.

In Gao Zhixi's seriation graves with the artefact complement ding-cauldron, he-bowl, hu-vase. and fang-vase are placed in the latest period because certain graves with this combination[6] can be securely dated to the late third century B.C.; and also because this combination is by far the most common in early Han graves in Changsha. These facts are certainly relevant. but they tell nothing about when the combination first came into use.

The combination ding-cauldron, dou-bowl, and hu-vase supposedly occurs only in period 4, but the same combination is found in two of the 27 early Western Han graves of the Academia Sinica excavations (Changsha 1957: 171, nos. 110 and 403). Both of these graves contain clay imitations of banliang coins. and therefore there is no question that they are later than 179 B.C. The combination ding-cauldron, dui-bowl, and hu-vase supposedly occurs only in periods 4 and 5, but in fact it occurs in three of the 27 early Western Han graves (Changsha 1957: 171. nos. 252. 332, 251); two of these graves also contain banliang coins. One lei-jar has been found in an early Han grave in Changsha (Li Zhengguang & Peng Qingye 1957: 66, no. A16). In fact all of the artefact types listed in table 2, except the li-cauldron, appear in early Han graves in Changsha. There may very well be considerable differences in style, but these are only incidentally mentioned by Gao Zhixi in his discussion. and they play no part in the formal description of his chronology.

---

In the above I have attempted to prove that the traditional means of seriating the Chu graves of Changsha are not to be relied upon. Assuming that the excavated material from the 2000-odd graves still exists, it should be possible for workers in Changsha to develop a new relative chronology on a sounder methodological basis. Such a seriation would probably involve (as in the Yutaishan seriation discussed in Section 5 below) first developing a stylistic sequence for each ceramic or bronze artefact type, then joining these sequences together by considering which styles co-occur in graves. This work can only be done by archaeologists who have direct access to the actual excavated material; therefore until the Changsha archaeologists have undertaken it we will have no usable chronology of the Chu graves of Changsha.

When a relative chronology has been developed, the next task will be to supply absolute dates in the sequence. This would seem to be a terribly difficult undertaking. There is simply not enough material available which can supply absolute dates.

At first sight the most useful material for absolute dating would seem to be several large, rich, and independently dateable tombs which have been excavated in Changsha. For example tomb no. 1 at Liuchengqiao, mentioned in Section 2 above, which is securely dateable to the fifth century B.C., contains ten ceramic li-cauldrons and ten ceramic ding-cauldrons. One might expect then to be able to date some styles of li-cauldron and ding-cauldron by comparison with these. Unfortunately the li-cauldrons in this tomb do not resemble the li-cauldrons from the smaller graves which are our primary concern here. They seem closer to northern styles (see KGXB 1972.1: 72 & pl. 1). Gao Zhixi (1979: 244) gives a number of examples of artefacts in dateable tombs which contradict his chronology. He concludes that the powerful persons who were buried in these tombs were conservative and held on longer to ancient traditions. The general point here is that different social classes are likely to have differed considerably in their religious practices, and that therefore their burials should not be mixed together in a single seriation .

In arguing for the antiquity of the li-cauldron in Changsha graves Gao Zhixi does not bring up the older arguments concerning plain mirrors and the Luo site which have been discussed in Section 2 above. His only argument now concerns the use of li-cauldrons in burials in north China. He first discusses the level of material culture north and south. and concludes that the south had a lower level in the Neolithic and Shang, but had reached the level of the north by the beginning of the Western Zhou period.

In view of these considerations. since the northern and southern cultures in the Zhou period each inherited the Shang cultural tradition and devcloped it separately. in cultural development they must have been basically concurrent. and their respective levels were not very different. There are merely small differences within a greater unity. Thus it is permissible to compare artefacts from the Changsha Chu graves with artefacts from Eastern Zhou graves in the Central Plains region, and thereby determine their dates. . . . In Luoyang the ceramic li-cauldron disappears by the late Spring and Autumn period: in the Fengxi 灃西 area [near Xi'an, Shaanxi] it disappears at the end of the Spring and Autumn or the beginning of the Warring States (these seem to be dated somewhat too late).[7] The disappearance of the li-cauldron must have been approximately simultaneous in north and south. Therefore the disappearance of ceramic li-cauldrons in Hunan . . . must also have been at the end of the Spring and Autumn period. (Gao Zhixi 1979: 242. 243)

I shall not here raise any objection to the statement that the level of material culture was approximately the same in the north and the south. The question I would ask is: what relevance does this broadly-conceived "level" have to specific burial practices? Gao Zhixi is perplexed by the different dates for the disappearance of the li-cauldron in two places (over 300 km apart) in the north and can only suggest that the Fengxi graves are misdated. It seems much more likely that religious practices were simply different in the two places, regardless of whatever may be known of their respective levels of material culture.

As noted in Section 2 above, the li-cauldrons found in Chu graves are very different in style from those found in the north, and numerous archaeologists in Changsha have noted that no direct comparison is possible. Su Bingqi (1982). in a short article filled with interesting ideas, proposes the term "Chu-style li-cauldron" 楚式鬲 (or "ding-style" 鼎式 or "jia-style" 斝) for these vessels. He suggests that their development can be traced back to certain jia-goblets and he-pitchers 盉 of the Hubei Longshan culture at the end of the third millennium B.C. (On the Hubei Neolithic cultures see e.g. Wang Jin 1980; treistman 1968.) Yu Weichao (1980: 10-11) proposes a somewhat different development, tracing the Chu-style li-cauldron back to the ding-cauldrons of the Hubei Longshan. These hypotheses do not contradict Chen Xianyi's (1980: 167-169) proposed derivation of the Chu-style li-cauldron from some aberrant li-cauldron styles found at the early Shang city-site at Panlongcheng 盤龍城 in Huangpi, Hubei; rather they can be seen as two possible extensions of the derivation further back in time.[8] If either hypothesis can be substantiated, and extended to other Chu ceramic types, we shall have one corner of an answer to the vexed question of the origin of Chu and its relation to Shang and Zhou. An implication would be that the slight resemblance of the Chu-style li-cauldron to the li-cauldrons of the north is fortuitous and that it would be wise, in the interest of clarity, to invent a new name for it.

---

Thus there appears to be no material available to permit an estimate of the period in which ceramic li-cauldrons were placed in graves in Changsha: this practice may well have started in the eighth or seventh century B.C., and it may well have continued nearly to the beginning of the Han.

A variant of the same argument is used concerning the lei-jar. The ceramic lei-jars are very similar to bronze lei-jars found in Luoyang which are dated to the middle Spring and Autumn period; therefore the ceramic lei-jars in Changsha should be fairly early. A single lei-jar co-occurs with a li-cauldron of style III, but mirrors found in graves containing lei-jars are of an allegedly later type than those found together with li-cauldrons. Therefore graves containing lei-jars must be slightly later than those containing li-cauldrons. The same objections raised above apply here.

Much the same type of argument is used for each of the periods in table 2. None leaves me with any real hope of determining absolute dates for any but the very latest periods of a future revised relative chronology.

5. The site of the Chu capital in Jiangling, Hubei

In the above I have limited my criticism of the traditional chronology of the Changsha graves to methodological considerations. It can also be compared with more reliably dated Chu archaeological material which has recently become available.

The site of Ying 郢, the capital of Chu, is about 5 km north of modern Jingzhou, the administrative centre of Jiangling County, Hubei. It was declared an Important Cultural Site of Hubei Province in 1956 and a National Important Cultural Site in 1961. Excavation planning and trial excavations began in 1965; at this time several large Chu tombs were also excavated outside the city walls (on these see Juliano 1972).

Excavations of the city-site itself began in 1973. They have been exceedingly cautious and thorough, and only a small part of the site has been uncovered. The results are only beginning to be published; the most important report to date is a two-part article published in 1982 (KGXB 1982.3: 325ff; 1982.4: 477ff).

The site is said to be the largest ancient city-site in south China. The area inside the tamped-earth walls is a rectangle measuring 4.5 x 3.6 km; this is larger than London in the mid-nineteenth century. Within the walls have been found remains of palaces, dwellings, bronze foundries, and ceramics workshops.

The walls seem to have been built at the end of the Spring and Autumn period, but the site was occupied much earlier, perhaps from the beginning of the Spring and Autumn. There is no doubt that this is the site of Ying, which was the capital of Chu from 689 B.C. The destruction of the city in 278 B.C., in the course of the Qin conquest, is clearly discernible in the archaeological material. Qin and Han graves have been found inside the city walls, and these constitute very sure evidence that the site was unoccupied.

---

Small graves which can be compared with the Changsha graves are found on nearly every hill or mountainside in the immediate vicinity of the city-site. Guo Dewei (1982: 157) estimates that there are over 20 cemeteries which contain more than 2000 graves each. The cemetery in which most graves so far have been excavated is at Yutaishan 雨泰山, about 1 km northeast of the city wall; a report on 558 graves excavated here has just become available (Jiangling 1984).

Two earlier attempts at a chronology of the Yutaishan graves (KG 1980.5: 391-402; Guo Dewei 1982) have now been superseded by the chronology given in the excavation report. Here a relative chronology is presented and argued in meticulous detail, and one may hope that it will serve as a model for all future work on Chu chronology. Its periodisation of ceramic artefact styles is summarised in Figure 7. The absolute dates for the six periods in this relative chronology are somewhat more uncertain. They are of necessity derived by comparison of artefacts with dated material from large rich tombs, and some of these are as far afield as Luoyang and Changsha. However it is known with some certainty that the city-site was occupied through the time-span covered by the chronology, and it seems reasonable to take the dates given as the best approximations which can at present be obtained:

1. Middle Spring and Autumn
2. Late Spring and Autumn
3. Early Warring States (ca. mid-5th century to early 4th century B.C.)
4. Early part of middle Warring States (early to mid 4th century B.C.)
5. Later part of middle Warring States (mid to late 4th century B.C.)
6. Early part of late Warring States (early 3rd century B.C., i.e. around the time of the Qin conquest in 278 B.C.) (Jiangling 1984: 134-139)

The dates proposed for periods 4-6 are based primarily on bronze artefact styles, and therefore are probably more reliable than those for earlier periods, which are based on ceramic styles. The Chu graves of the Jiangling area are more numerous, better preserved, and generally richer than those of Changsha, and it is not unreasonable to expect that future excavations will provide material for a more reliable absolute dating.

In comparing these graves with the Changsha graves we may start with the grave-forms. Figure 8 gives histograms for L/W, the ratio of length to width at the bottom, for graves of each of the six periods; these should be compared with the histograms given in Figure 5 for the Changsha graves. The first aspect which strikes the eye is the difference between the two sets of histograms. The Changsha histograms have clear peaks at L/W = 1.54, 2.37, and 3.46. The two most obvious peaks in Figure 8 are at about 2.0 in period 4 and about 2.5 in period 2. Long narrow graves corresponding to group 1.1 in Changsha, with L/W > 3 occur in all periods except the earliest. There may be a slight peak at about 1.5, corresponding to groups 2+3, in period 6, but this is not very clear.

In the Changsha material the grave-shapes divide neatly into three approximately normally-distributed groups, but in the Jiangling material the situation is much more complex. Sufficient data is available for a more sophisticated statistical treatment, but these will require more advanced data processing equipment than the slips of paper and pocket calculator which I have used in preparing Figures 5 and 8.

The conclusion concerning grave forms seems only to be that, with respect to the one parameter L/W, the Changsha graves are rather different from the Jiangling graves, and a simple comparison would be hazardous. At any rate there is no support here for the assumption that narrow graves are earlier than others.

---

Turning to the ceramic styles found in the Jiangling graves (Figure 7) several similarities and differences can be seen in comparison with Gao Zhixi's proposed chronology. The ding-cauldron, fang-vase, and he-bowl (columns 5 and 8, 10, and 11) occur in approximately the same absolute time-spans in both chronologies; but the li-cauldron, bo-bowl and guan-jar (columns 1, 2, and 3) appear much later in the Jiangling chronology than in Gao Zhixi's. Furthermore li-cauldrons and ding-cauldrons overlap considerably in time. Fu-dishes (column 6), as far as I know, do not occur in Changsha graves. Flat-topped dui-bowls, placed by Gao Zhixi in period 4, seem not to occur in Jiangling Lei-jars (not listed in Figure 7) occur in periods 2-6 (e.g. grave nos. 79, 142, 183, 176, 555).

Since mirrors have sometimes been considered to have chronological significance in Changsha it is interesting to see that the nine mirrors found in the Yutaishan graves are all plain and undecorated. One of these is from period 4, two from period 5, and four from period 6, two are from undateable graves.

It has been seen above (Section 4) that Gao Zhixi's chronology depends heavily on comparison with northern material. Jiangling is much closer to Changsha than Luoyang or Xi'an, and the excavated material definitely represents Chu culture. The many differences between Gao Zhixi's chronology and the Jiangling chronology therefore constitute an additional argument against Gao Zhixi's method.

It would be tempting to make a simple comparison between the Jiangling and Changsha material, and date the Changsha graves according to the more secure dates for the ceramic types found in Jiangling. This would however be quite wrong. Though we know next to nothing about life in ancient Changsha, it is a reasonable guess that it was a sleepy provincial settlement where fashions changed much more slowly than in the capital. A number of empirical studies in archaeological method have shown that clearly related stylistic sequences, separated by only short distances, can be considerably displaced in time; so that a stylistic change which occurs at a specific time in one place occurs at an earlier or later time in a place only a few km away (see e.g. Deetz & Dethlefsen 1965; Dethlefsen & Deetz 1966; Sharer & Ashmore 1979: 366-367).

An example from later Chinese history can exemplify this phenomenon. In A.D. 317 Luoyang, the capital of the Jin Dynasty, fell to nomadic invaders; this completed their invasion of north China, and the ruling house of Jin moved its capital south to the area of modern Nanjing. This area had been a part of the Han empire, and at least nominally under the rule of Jin, but had always enjoyed considerable social and political autonomy. Settlers from north China had trickled in since the beginning of the Han. The aristocratic families of the region had maintained a culture which, while not lacking in purely local elements, largely reflected a northern culture of the Han period which in the north had changed considerably in the intervening centuries.

The transfer of the capital in A.D. 317 brought an influx of powerful northern families who immediately became the socially and politically dominant group in the region. They brought with them new ideas and religious usages, and their dominant position meant that many southerners immediately attempted to assimilate to the new group. Others stoutly maintained the older traditions: Ge Hong (284-364) wrote angrily and ironically of the decadence of the northerners, especially concerning their lax view of mourning and the burial of the dead (Tang Changru 1955).

A study of fourth-century A.D. graves in the Nanjing area would probably reveal two rather different grave-types, reflecting the burial practices of the two segments of the population. On the other hand, if the dating methods used in Changsha were applied uncritically to this material the graves of the conservative native aristocrats might be mistakenly dated to the Han period, several centuries too early.

A plausible hypothesis inspired by this story, or rather a possibility which must be borne in mind, is that the earliest settlers in the Changsha area, who may have arrived from the Jiangling area as early as the seventh century B.C., brought traditions with them which were maintained in Changsha long after they had changed in Jiangling. The use of li-cauldrons is an example. This dies out in Jiangling by the end of the fourth century B.C., but "old families" in Changsha, descendants of very early settlers, might have maintained the tradition long after, in spite of or in defiance of new practices brought to the area by later settlers.

6. Iron in Jiangling and Changsha

In an important article Huang Zhanyue (1976) has brought some order into the question of the date of the first use of iron in China. Using rigorous philological criteria he rejects, one by one, all of the arguments based on written sources. Considering then the archaeological evidence, he rejects several arguments on various grounds: methodological confusion, incorrect dating, or inadequate reporting of excavation details. He concludes that the earliest reliably dated iron artefacts so far found and published in China come from certain graves in Hunan and Jiangsu, and that all of these are to be dated ca. 500 B.C. [9] Figure 4 shows some of these artefacts.

The two graves in Luhe 六合, Jiangsu seem to be quite reliably dated, on the basis of bronze styles and inscriptions. The graves in Hunan, however, are dated to ca. 500 B.C. because they contain li-cauldrons; that is, Huang Zhanyue accepts the traditional chronology for Chu graves in Hunan which I have attempted to refute above. I claim that these artefacts cannot in fact be dated more precisely than "pre-Han".

Gao Zhixi (1980: 59; see also Hou Dejun 1984) gives some newer data from unpublished excavations: 261 iron artefacts have been found in something on the order of 1000 Chu graves in Changsha; 20 of these artefacts would be dated by his chronology to the late Spring and Autumn period.[10] Looking to the more reliably dated material from Jiangling, the situation looks rather different. In the 558 graves excavated at Yutaishan the iron artefacts found were as follows:

Period 4:
7 bronze ding-cauldrons with cast-iron legs
2 crossbow-bolts with bronze tips and iron shafts
Period 5:
1 bronze ding-cauldron with cast-iron legs
1 cast-iron spade-cap (similar to Figure 4.3)
1 cast-iron axehead

(It is extremely interesting that the two earliest applications of iron seen here do not exploit the superior mechanical properties of iron. Perhaps iron was used for the ding-cauldron legs to give a decorative effect, and for the crossbow-bolt shafts because it was cheaper. The number of iron artefacts found is still small, however, and finds from future excavations may change this picture entirely.)

Table 3. bronze and iron artefacts of comparable function in four reports of excavations of Chu graves.

Artefact type	Changsha 61 graves (Li & Peng 1957)	Changsha 73 graves (Changsha 1957)	Changsha 209 graves (Wen Daoyi 1959)	Total Changsha 343 graves	Jiangling 558 graves (Jiangling 1984)
iron digging implements	1	1	3	5	1
bronze digging implements	0	0	0	0	0
iron axehead	0	0	3	3	1
bronze axehead	0	0	0	0	0
total implements	1	1	6	8	2
iron crossbow-bolt	0	2	ca. 2	ca. 4	0
bronze-iron crossbow-bolt	0	8	ca. 2	ca. 10	2
bronze crossbow-bolt	8	14	ca. 12	ca. 34	183
total crossbow-bolts	8	24	ca. 16	ca. 48	185
iron sword	3	0	7	10	0
bronze sword	8	21	82	111	172
bronze dagger	0	0	0	0	40
total swords and daggers	11	21	89	121	212
iron ji halberd-head	0	0	1	1	0
bronze ji halberd-head	0	0	0	0	7
bronze spearhead	4	6	15	25	15
bronze ge dagger-axe head	3	5	12	20	97
total shafted weapons	7	11	28	46	119
iron writing-knife	0	0	7	7	0
bronze writing-knife	0	0	2	2	15
iron scraper-blade	0	0	5	5	0
bronze scraper-blade	0	0	0	0	0
total knives	0	0	14	14	15
iron ding-cauldron	0	0	0	0	0
bronze-iron ding-cauldron	0	2	some	ca. 3	8
bronze ding-cauldron	0	4	10	14	10
total ding-cauldrons	0	6	ca. 11	ca. 17	18
iron ring	0	1	0	1	0
bronze ring	0	1	7	8	0
total rings	0	2	7	9	0

It can be seen that surprisingly few iron artefacts have been found in the Yutaishan graves as compared with the 261 found in similar graves in Changsha. table 3 presents a more rigorous comparison, giving data on iron artefacts and bronze artefacts of comparable function found in Chu graves in Changsha and Jiangling. These data may be summarised in the following table, which gives the percentage of artefacts which are of iron or a combination of bronze and iron.

	Changsha	Jiangling
implements	100	100
crossbow-bolts	29	1
swords and daggers	8	0
shafted weapons	2	0
knives	86	0
ding-cauldrons	18	44
all types in table 3	19	2

These figures appear to suggest that iron was used much more in Changsha than in Jiangling. This need not necessarily be a surprise: it is an often-remarked phenomenon world-wide that the earliest use of iron tends to occur at the periphery of a high culture rather than at its centre. In the beginning, iron is a cheap and inferior substitute for bronze, and only after it has been in use for some time are the techniques developed which make it a superior metal for most applications. It is certainly a tenable hypothesis that this development took place in Changsha.

Other explanations of the figures above are also possible. Local peculiarities in choices of grave-goods may have played a part. Iron digging implements may have been more necessary in the heavy clay soil of Changsha, and in this soil they may also more easily have become detached from their wooden handles and lost in the grave-fill.

Nevertheless a simple and apparently incontrovertible chronological fact overshadows these considerations. The Yutaishan graves range in date from the seventh century B.C. to 278 B.C., when the city was destroyed by Qin. On the other hand the archaeological record does not reveal any disturbance of life in ancient Changsha at the time of the Qin conquest (Changsha 1957: 163), and the graves classified as pre-Han range in date from some indeterminable time to shortly after 179 B.C. (Section 2 above). Thus a later period of at least a century is included in the Changsha material, and this seems to be the most likely explanation for the fact that more iron is found in the Changsha graves.

From the Yutaishan material it seems that iron in Chu cannot be taken back before the fourth century B.C. In the city-site itself a few more than 11 iron implements have been found (KGXB 1982.3: 334, 341, 347; 1982.4: 477, 483, 490, 495). One of these, a cast-iron spade-cap, was found in a stratum which has been dated to the late Spring and Autumn period (KGXB 1982.3: 347-348). However this date is arrived at by comparisons of ceramic artefacts with the Yutaishan sequence, and some of these comparisons, to my eye at least, seem to be of rather doubtful validity. In particular a bo-bowl found in the stratum (KGXB 1982.3: 346, fig. 14 no. 1) is strikingly like the bo-bowl of period 5 of the Yutaishan sequence. It would be wisest to wait for a more detailed excavation report before making too much of this single find.

Another Chu iron artefact of great potential importance is a dagger with jade stem found in a Chu tomb at Xiasi in Xichuan County, Henan 淅川縣下寺, 300 km north of Jiangling. This is one of nine large Chu tombs excavated here. One of these (M2) has been identified on the basis of bronze inscriptions as the tomb of Zigeng 子庚, a Chu prince who died in 552 B.C. (Zhang Jian & Zhao Shigang 1980; Zhang Jian 1980; Liu Shijin et al. 1981; Gu Tiefu 1985; Zhao Shigang 1986).[11] The dagger was found in tomb M10. The published excavation reports concentrate on Ml and M2, and very little information is available as yet on M10. Zhang Jian (1980: 23) places it in the late Spring and Autumn period, but this dating is based on two unargued assumptions: (l) All of the tombs are from the Spring and Autumn period. (2) The placement of the tombs, in a north-south row, has chronological significance, with the northernmost tombs latest. M10 is the next-to-last in the row, and therefore is late Spring and Autumn. Other scholars appear to disagree with the second assumption (Liu Shijin et al. 1981: 127; Gu Tiefu 1985: 82). Furthermore, since the tombs were found at the edge of a modern reservoir, it is not clear that the apparent arrangement in a row is real: it is possible that we have here the western edge of a larger cemetery which is now submerged (cf. Wen Bigui 1983: 542).

The claim that all of the nine tombs are from the Spring and Autumn period cannot be tested without more details on the excavated material. The only information available on M10 is that there were among its grave-goods four bronze ding-cauldrons (Zhang Jian 1980: 21), one bronze dui-bowl, and one bronze belt-hook (Zhang Jian & Zhao Shigang 1980: 16, 19). The dui-bowl and belt-hook are the only artefacts of these types found in the nine tombs. Wang Renxiang's exhaustive study of ancient belt-hooks (1985) indicates that the belt-hook did not become popular until the Warring States period. He lists only 17 belt-hooks of the Spring and Autumn period, of which only four were found in Chu contexts. One of the four is the one under discussion here, and two are from graves in Shaoshan, Hunan, which are dated by the traditional Changsha chronology which I have attempted to refute above (Zhou Shirong 1977). Considering only this one artefact, therefore, it is possible but far from certain that tomb M10 is from the Spring and Autumn period. When more has been published about this tomb and its grave-goods a definite statement may be possible.

7. Iron in the state of Wu

The earliest reliably dated iron artefacts so far found in China are therefore the two found in graves of about 500 B.C. in Luhe, Jiangsu (Figure 4.4-5). This is at the northern extremity of the ancient state of Wu 吳, which covered southern Jiangsu, southeastern Anhui, and northern Zhejiang. Wu was a barbarian state, and its "tumulus" graves are very different from those of north China or of Chu. A very useful article by Zou Houben (1982) summarises the currently available material on these graves. Something on the order of 100 Wu tumulus graves have been scientifically excavated, and bronzes were found in only three: five vessels and two ge dagger-axes (Zou Houben 1982: 67). On the other hand, when chance finds are taken into account, the total number of bronze artefacts with known provenance is over 240, from 18 graves and 7 buried hoards (Liu Xing 1981a: 28). The artefacts include vessels, musical instruments, weapons, agricultural implements, and copper ingots. A disturbing implication emerges here: apparently the modernisation of industry and agriculture in this region in the last decade or two has led to the destruction of large numbers (hundreds) of tumuli; and only those which happened to contain bronze artefacts have seemed to the workers to be worth reporting to archaeologists. This loss of precious data is in any case lamentable, but there is a specific problem with respect to our present concern: a badly rusted piece of iron might not be recognised as a valuable artefact, or even as an artefact at all, by persons without archaeological training. The only iron artefacts so far reported from Wu tumulus graves are parts of bronze artefacts. These are two iron rings (believed to be the remains of a casting core) inside a bronze yong-bell 甬鐘 and iron cores inside the handles of two zheng-bells 鉦. The three bells were found in Liyang, Jiangsu. They are chance finds, and therefore very difficult to date: on stylistic grounds they are considered to be later than certain other bells dated to the late Spring and Autumn period (Liu Xing 1981b: 108, 109).

Two other Wu iron artefacts, said to be from the Spring and Autumn period, are noted very briefly by Zou Houben (1984). These were found in the excavation in 1983 of twenty-four "stone chamber tumulus graves" at Wufengshan 五峰山 in Wuxian County, Jiangsu, 17 km southwest of Suzhou. We must wait for the excavation report before drawing conclusions on this find.

Wu archaeology is still in its infancy, and the same might well be said of the archaeology of Chu. Future excavations in both regions, and the detailed publication of past excavations, may be expected to bring greater clarity into the picture of iron usage presented above. What seems to be clear now is that iron was probably used in both Wu and Chu before it was used in the north. The best evidence available at the moment suggests that iron was first used in Wu, but this evidence is not yet very strong.

Future work on this question should also focus on the bronze industries of these regions. Where do we find a bronze-based technology which would be likely to provide favourable conditions for the innovation of iron? Here I would point out especially the use of bronze agricultural implements. These are extremely rare in north China, and somewhat more common in the south and southwest. They occur most commonly by far in the ancient territory of Wu (e.g. Liao Zhihao 1982; Liu Xing 1982; 1985; Chen Zhenzhong 1982; Lu Maocun 1984). A widespread use of bronze agricultural implements might be expected to lead to a demand for something cheaper, and thus to the use of iron. In weaponry, on the other hand, there would have been much more resistance to the introduction of a cheap and inferior substitute for bronze.

I have elsewhere suggested a hypothesis concerning the technical details of how the innovation of iron may have occurred in China (Wagner 1986: 7-10). The copper-smelting furnaces of the sixth century B.C. excavated at Tonglüshan 銅綠山 in Daye, Hubei smelted rich copper ores and used iron ore as a flux. Much more iron than copper went into these furnaces, and in normal operation it came out again as Wustite (FeO) in the slag. If too much fuel were to be charged in such a furnace, or if the air-blast was not strong enough, the atmosphere in the furnace would become excessively reducing, and cast iron would be obtained instead of copper. It is therefore likely that iron was well-known to the copper-smelters as a sign of improper furnace operation. The discovery of iron was, under this hypothesis, nothing more than the discovery that this unwanted by-product could be useful.

Where might this discovery have occurred? There is no sign that iron was ever smelted at Tonglüshan in ancient times; and there would have been very little reason for it at this fantastically rich copper mine. In an area in which copper ores were not plentiful there would have been a much greater economic incentive to substitute iron for bronze. The territory of Wu is one such area: the copper deposits in Jiangsu are all very small, and are usually found together with much larger deposits of iron (Shan Shumo et al. 1980: 19; C. C. Liu & J. C. Chao 1924, Ch. text p. 71).

The research reported here is part of a project on iron and steel in ancient China supported by the Danish Research Council for the Humanities, the Carlsberg Foundation, the University of Copenhagen, and Dr. Joseph Needham. Harish Gaonkar kindly drafted Figures 5, 6, and 8. Thanks are due especially to several colleagues whose criticism of an earlier draft led to extensive revision of this article: Bennet bronson, Albert E. Dien, Nancy Price, Bent Sørensen, Per Sørensen, Henrik Tauber, and Olfert Voss. None can be held responsible for shortcomings that remain. Postscript, April 1987: This article was complete in essentially its present form in December 1985. In preparing it for publication in Acta Orientalia I have made no major changes, but have added references to a few matters which have been published, or have come to my attention, since then.

8. References and abbreviations

Works are generally cited here by author-name and year of publication. In the case of anonymous works and works by corporative authors, articles are cited by an abbreviation of the journal title and books by one or two key words from the title. Translations of titles are normally my own; title translations taken from the work in question are placed in quotation marks.

Notes