Review in Journal of the economic and social history of the Orient (Leiden), 1993, 36: 304-308
Donald B. Wagner
This book will receive a warm welcome from historians who believe, as I do, that technology is an inseparable part of history. It provides three different aids to such historians. First, it is a kind of textbook of archaeometallurgy, useful especially for those who already know something of modern metallurgy. Second, throughout its exposition it gives interesting and well-researched historical examples. Finally, it includes chapters which take up broader historical problems and shows that serious technical study can give decisive insights in their study. Readers of JESHO will be especially pleased to know that this book, unlike most histories of technology, takes Asia seriously, drawing a significant proportion of its examples from contexts in Asian history. In dealing with China the authors have even used a certain amount of Chinese-language material.
In this review I shall consider only the book's final chapter, "Economics and geography". It draws on the foregoing technical discussions in a fascinating discussion of the complex interplay of factors which has determined where and how iron industries and the associated trade routes developed through history in different parts of the world. The authors distinguish between "commodity irons" and "specialty irons" (such as various kinds of steel); here I shall confine myself to their discussion of commodity irons.
Rostoker and Bronson stress two technical factors as important for this discussion: Iron is cheap, but still valuable enough to repay transportation for a certain distance, and it is of wide occurrence, with small quantities of ore virtually everywhere and enormous quantities in many places. One implication of these facts is that the formation of monopolies or oligolopies, or the establishment of centralised political control, should be virtually impossible. Another is that profits will be low in such an industry: in the language of economic theory, there will be a strong tendency to perfect competition, so that profits tend to zero.
The mid-eighteenth through mid-twentieth centuries seem in fact to be a historical anomaly: a period in which the making of ordinary iron was profitable and, briefly at least, at the forefront of the development of materials technology. Otherwise, except for several periods in Chinese history there are few indications that the making of commodity iron was regarded as exciting. The smelting and fabrication of iron have been by and large conservative, routine, and small scale. [pp. 186-7]And the dismal condition of the world steel industry today is in fact the long-run norm. As this passage hints, the authors see high profits in iron production as the result of continuous innovation: and large-scale production as a result rather than a cause of high profits.
In considering the geographical distribution of the iron industry, Rostoker and Bronson start by showing that a uniform distribution of iron production throughout the inhabited world would be possible; that is, all iron used in a pre-industrial economy could be produced on a village-by-village basis, with no trade at all. But the only example they find of such a phenomenon is in a part of Borneo; the historical norm is uneven distribution, and they conclude: "As far as is now known, the great bulk of early iron was smelted in a minority of settlements and supplied to the rest through trade and other modes of exchange." (p. 180)
The authors explain the development of iron production centres and trade in pre-modern economies as due to the inherent geographical variation in production costs, caused either by differences in natural endowments or by local improvements in the technology. Here an additional element, barely touched on by the authors, may be added. The relevant cost is opportunity cost, and depends on what other uses exist in the given locality for the labour and capital used in iron production. Actual differences in endowment, or in efficiency in the use of production factors, need not be involved. For example, the eighteenth-century Norwegian writer Ole Evenstad blamed the decline of bloomery iron smelting in his own time on the timber trade: the peasants could use their slack-season labour to greater benefit in cutting timber for sale than in producing iron for local use.
There is then a discussion of trade networks and the numbers of network nodes to be expected in particular circumstances, backed up by numerous historical examples. In an undeveloped economy we should expect a very large number of small "locally oriented" iron-smelting centres; in a partially-developed economy, "like those of India and West Africa during the seventeenth and eighteenth centuries or Europe in the Roman period and Renaissance" (p. 183), there will also be medium-sized "subregionally oriented" centres. (Note here that medieval Europe is considered to be an undeveloped economy - I think I disagree.) As for developed economies:
Since only two examples of a developed pre-industrial economy exist - China at various periods of its history and Europe during the sixteenth to eighteenth centuries - model building is simplified. . . . While locally oriented small producers will die out, medium-sized makers of commodity iron and specialty producers will survive everywhere except in the immediate hinterland of large cities and industrial centers, perhaps with little change in scale or technology. In the metropolitan areas, however, [as in eighteenth-century France, a new] type of producer appears, . . . oriented toward grande commerce and with markets as large as entire regions. Size here is economical: hence large furnaces and large integrated enterprises begin to appear. The low price of the iron produced, as well as the manufacturers' ability to coerce traders and government regulators, hastens the demise of small local ironworks. In a few places competition among large and medium-sized firms might be intense enough to lower the price of iron to a point where consumption in the household and the farm begins to rise. [p. 184]Here the authors argue strongly against the widespread assumption that the blast furnace provides significant economies of scale. The "size" in this passage is the size of the enterprise, not of its furnaces, and the advantages which large size provides are organisational or political rather than technical.
The examples of pre-industrial iron production in a developed economy which Rostoker and Bronson see in Chinese history are in the Han period (206 B.C. - A.D. 220) and the Song period (960-1279), and in the province of Shanxi in the nineteenth and early twentieth centuries. Here they run into some trouble, for their sources are often inadequate. They state for example: "In economically developed societies, like Roman Italy and Han China, small producers using minor [ore] deposits mostly ceased operations . . . ". They cite no source for this; I do not know what sources may exist for Roman Italy, but for Han China I doubt that any source exists which can prove or disprove the statement.
Rostoker and Bronson follow Robert Hartwell in stating that the iron industry in the Song period reached a peak in per capita production which was never reached again until modern times. Hartwell has made great contributions to the economic history of the Song period, and Western economic historians appreciate his work because he gives precise numbers: for example Chinese iron production in the year 1078 was 125,000 short tons. But Chinese and Japanese historians have shown that these figures are based on an oversimplified interpretation of the relevant sources. A proper interpretation of the sources will require a thoroughgoing study of the administrative apparatus which produced them, work which has barely begun. And for later periods the available sources are not nearly as useful as those for the Song, for in the Ming and Qing periods taxes on iron production were normally collected on a per-furnace basis rather than as a percentage of production, and bureaucrats no longer had a need to know how much was actually being produced.
With respect to Shanxi in recent centuries it is important to notice that, in the period for which we have good sources, the industry was in a deep decline because of foreign competition, and that this probably had an effect on the technology used. Foreign competition had depressed prices drastically, and the situation was made even worse by the fact that the Shanxi iron producers could (or would) compete only in price, not in quality. In 1870 Ferdinand von Richthofen saw signs of decline everywhere in Shanxi, but estimated that annual production in the entire province was 160,000 tons; thirty years later, William H. Shockley's estimate was only 50,000 tons. Analyses by T. T. Read of pig iron from Shanxi show such large amounts of sulphur in the iron (lowest 0.12%, highest 0.61%) that it must have been extremely difficult to use, and can only have been sold at very low prices. It is quite likely that, in the days before foreign competition depressed prices, the technology of iron production in Shanxi was better and produced a better quality of iron. Experiments with essentially the same "crucible smelting" process in Höganäs, Sweden, showed that the sulphur content could be reduced to 0.01-0.03% by the addition of a small amount of limestone to the crucible charge combined with careful temperature control. The same could easily have been done in Shanxi in better times, or perhaps other tricks were known and used.
In an appendix (pp. 205-206) the authors give a brief appreciation of "the first American metallurgist", Frederick Overman (1803-52). Comparing him with the great English metallurgist John Percy (1817-89), they write: "Both had impressive analytical abilities, but Overman comes to strong positive, sometimes dogmatic conclusions appropriate to the demands of decision-making whereas Percy very often took an ambiguous position as befits one who was conscious that the facts were not all in place . . . " This evaluation of Overman could be used without modification as an epitaph for William Rostoker (who died in 1990). There is an enormous amount of good sense in his published work, often based on careful empirical or theoretical work: one of many examples in this book is his experiments on the effect of sulphur on the surface tension of liquid cast iron (p. 22, fn. 9). On the other hand he often made ex cathedra statements purely on the basis of his long experience as an engineer. Throughout this book there are examples of such statements: undoubtedly most are correct, but all must be checked out empirically or theoretically before they can be accepted. One example is the statement on p. 192 that, in the carburization of iron, "The influence of the CO/CO2 ratio is independent of dilution by a neutral gas such as nitrogen." This is not correct:, as as can be confirmed by reference to any textbook on engineering thermodynamics.
) E.g. Qi Xia, Song dai jingji shi, vol. 2 (Shanghai Renmin Chubanshe, 1988), p. 555; Yoshida Mitsukuni, Chugoku kagaku gijutsu shi ronshu (Tokyo: Nihon Hoso Shuppan Kyokai, 1972), pp. 364-369.
) Ferdinand von Richthofen, Baron Richthofen's letters, 1870-1872 [to the Shanghai Chamber of Commerce] (Shanghai: North China Herald, n.d.), p 38; William H. Shockley, "Notes on the coal- and iron-fields of southeastern Shansi, China" (Transactions of the American Institute of Mining Engineers, 1904, 34: 841-871), p. 870; Thomas T. Read, "The mineral production and resources of China" (Transactions of the American Institute of Mining Engineers, 1913, 43: 3-53), p. 27.
) Emil Sieurin, "Höganäs järnsvamp" (Jernkontorets annaler, 1911, 448-493), pp. 452, 458-459.
) Wang Zhongshu, Han civilization (New Haven & London: Yale University Press, 1982), p. 125; Han dai kaoguxue gailun (Beijing: Zhonghua Shuju, 1984), p. 68.