THE EVIDENCE FOR THE HD

The very notion that in eighth-century Japan one million pieces of text were printed on paper and one million miniature pagodas manufactured to contain them seems improbable both in terms of logistics and of technology, and perhaps it is for that reason that most histories of the Nara period ignore the HD altogether, although they pay much attention to other contemporary grandiose projects, especially the construction of the Great Buddha at the Tōdaiji in Nara.Footnote ³ Yet the official record is clear: it is to be found in the second of the six official histories of Japan, the Shoku nihongi 續日本紀, which was completed in 797 and which records as follows in an entry for the year 770:

In view of the significance subsequently attached to this event, the official record is laconic, and it does not even mention that the dhāraṇī were printed.

What is missing from the Shoku nihongi account is partly supplied by a short passage in the Tōdaiji yōroku 東大寺要録. This compilation of early records relating to the Tōdaiji temple in Nara bears a preface dated Kajō 嘉承 1 [1106], that is to say more than three hundred years after the HD; unfortunately, the compilers make no mention of the records they drew upon. The Tōdaiji yōroku includes an account of some structures at the Tōdaiji called the Tōzai shōtōin 東西小塔院, which appear from their names to have been two buildings housing miniature pagodas, and this adds some further details:

This passage identifies the sūtra from which the spells were taken by the title of its Chinese translation, namely Wugou jingguang ta tuoluoni jing 無垢浄光大陀羅尼經 (J. Muku jōkō dai darani kyō; hereafter simply M), meaning “sūtra of the dhāraṇī of pure unsullied light”.Footnote ⁶ However, since the efficacy of dhāraṇī depended upon their oral qualities, when sūtras containing them were translated into Chinese or Tibetan the dhāraṇī themselves were not translated but transcribed phonetically. Thus the spells that form part of the HD cannot be said to be in any form of the Chinese language and simply make use of the Chinese script.Footnote ⁷

The passage in the Tōdaiji yōroku states explicitly that the spells were printed, and refers in the final note to the existence of an oral tradition associating the whole exercise with atonement, presumably on the part of Shōtoku, for the blood spilt when forces of the state crushed the rebellion launched in 764 by Fujiwara no Nakamaro 藤原中万呂 (also known as Emi no Oshikatsu). It also says something of the ultimate fate of the HD and it implies, but does not state, that the two Small-Pagoda Pavilions erected at the Tōdaiji were built to house its share of the HD.

More than 45,000 miniature pagodas and 3,962 printed dhāraṇī survive in the Hōryūji temple outside Nara.Footnote ⁸ It is not known how many survive in total: they have been passing into private hands at least since the 1790s.Footnote ⁹ In 1877 the designer Christopher Dresser was offered a pagoda with a dhāraṇī, and in 1881 Ernest Satow reported that specimens of HD were available from dealers in antiquities; meanwhile, some of the HD from the Hōryūji had been presented to the Imperial Household in 1878 and 3,000 were put on sale in 1908 to cover the Hōryūji's restoration expenses.Footnote ¹⁰ As a result, there are large numbers now to be found not only in various Japanese libraries but also in various collections outside Japan.Footnote ¹¹ Nevertheless, the total number of pagodas known to be extant still falls short of 50,000.

The dhāraṇī or spells were printed on long, narrow pieces of paper which in size and shape resemble the tala-leaf manuscripts of South and South-east Asia. It is possible that such manuscripts may have been familiar in Japan in the Nara period: an example dating from the seventh or eighth century and containing part of an incantation in Sanskrit was long preserved in the Hōryūji.Footnote ¹² Again, the total number of printed spells extant around the world is unclear, though it is unlikely to be more than 20,000.

The point of greatest dispute in Japan is the method used to print the spells, for the debate focuses upon the HD as printing history and consequently neglects other questions.Footnote ¹³ Although it might seem obvious that they must have been printed with woodblocks, the view that they were printed with metal plates, first articulated in 1795, still has much support.Footnote ¹⁴ At a seminar in 1965 it was stated that in some cases wood grain had been identified in the impressions left by the ink and that experiments using metal plates did not produce the same effects as the surviving printed spells.Footnote ¹⁵ The tentative conclusion reached was in favour of xylography, but Nakane in Reference Nakane Katsu1987 was not so sure, finding that some spells seemed to have been printed with woodblocks and others with metal plates.Footnote ¹⁶

Each of the spells survives in at least two versions, a “long” and a “short” version, which has led some to suppose that each of the eight plates or woodblocks would have been required to print the large number of 125,000 copies. This supposition, based solely on the spells that have survived from the Hōryūji, is suspect. The extant Hōryūji pagodas were all produced in 767–768, suggesting that those for other temples were produced separately, either earlier or later; since we lack spells originating from the other temples we have no way of knowing whether they were printed from the same woodblocks or metal plates. It clearly possible that different woodblocks or metal plates were used for the spells consigned to other temples.Footnote ¹⁷

The production of metal printing plates would not have been beyond the abilities of craftsmen in Nara, for the Wadō-kaichin 和同開珎 coins were cast in 708; however, although archaeological research has identified the site where these coins were cast, no such evidence has come to light to support the metal-plate theory for the printing of the HD.Footnote ¹⁸ Experiments carried out in recent years with metal plates have convinced some that this was how the HD were printed.Footnote ¹⁹ Experiments, however, are one thing, and hard evidence is another, and of the latter there is as yet none to be had.

Many of the pagodas carry handwritten inscriptions on the underside of the base and these include dates from the late 760s. The conclusion is inescapable that the documentary accounts are accurate and that the pagodas were manufactured in the eighth century. The printed spells themselves are undated but if the pagodas were manufactured to contain them, then the inevitable conclusion is that they, too, were printed in the eighth century. Although it is the pagodas that are dated, not the spells, it is universally accepted that the spells were printed in the eighth century, and there are no good grounds for doubting this. So the question of the origins of printing in Japan and the reasons for undertaking printing on this scale beg to be answered. I take it as axiomatic that this was not the first time that printing had been practised in East Asia, and consider it unlikely to have been an indigenous development in Japan, so the HD need to be understood in a broader context which takes in other parts of East Asia beyond Japan.Footnote ²⁰

THE DHĀRAṆĪ SŪTRA FOUND IN KOREA AND THE DEBATE ON THE ORIGINS OF PRINTING

The main rival to the HD in terms of antiquity is a scroll containing the full text of M. It consists of the very same text from which extracts were taken for the HD and it was found inside a pagoda; what is more, it was indubitably printed xylographically. Its bearing upon the HD is obvious, but the connections are obscure because of the controversy that surrounds this find, and it will therefore be necessary to try to clear away some of the chaff.

The scroll, together with some reliquary objects and wooden miniature pagodas, was found in 1966 inside the Sŏkkat'ap 釋迦塔 stone pagoda of the Pulguksa 佛國寺 temple in Kyŏngju.Footnote ²¹ It carries no internal date and there are no extant documentary references to printing in eighth-century Korea, but it has been thought to date from some time before 751, on several grounds. First, the Pulguksa records date the temple's construction to 751; second, the other objects found inside the pagoda are of eighth-century Silla provenance and not of Chinese origin; and third, the text includes four of the unorthodox characters introduced by empress Wu and other variants that can be attested from eighth-century Korean and Japanese sources.Footnote ²² A fourth argument made for an early date is the narrow format of the paper which it shares with the HD and which resembles the shape of tala-leaf manuscripts; by the time that the copy of the Diamond sūtra, which was found at Dunhuang and is now in the British Library, was printed in 868, a larger format had become standard.Footnote ²³

Cumulatively these arguments are impressive, but they are not conclusive. First, the date of the construction of the pagoda does not guarantee that the contents were inserted at the same time.Footnote ²⁴ Second, the special characters invented by empress Wu were used well into the ninth century. This is evident from their use in Dunhuang documents, particularly when making an authentic copy of an older document in which they were used: since the sūtra M was translated in Wu's time, if it contained her special characters it would not be surprising to find them retained in subsequent copies or printed versions, and they are no reliable guide to dating.Footnote ²⁵ Finally, the argument from the format is not conclusive, for copies of the Baoqieyin tuoluoni jing 寶篋印陀羅尼經 printed in China in 975 and in Korea in 1007 share the same narrow format.Footnote ²⁶ Thus the date of the Pulguksa sūtra is open to argument.

The Pulguksa find has aroused mixed reactions. Korean scholars have suggested that the paper and the calligraphy were Korean and therefore that it was printed in Korea, and a Taiwanese scholar concurs.Footnote ²⁷ Elsewhere, however, doubts have been raised. Some Japanese scholars have cast doubt upon the accuracy of the Pulguksa records, and the common assumption in both Japan and China is that the scroll was printed not in Korea but in China.Footnote ²⁸ Many Chinese scholars take it as given that it was printed in Tang China, and Pan Jixing points out, rightly enough, that there are no documentary mentions of printing in Korea until many hundreds of years later.Footnote ²⁹ But this point applies equally well to Japan, to say nothing of China itself; and there is in fact documentary evidence of printing in late ninth-century Korea.Footnote ³⁰ In his most recent writing on the subject, Pan simply declares that the Korean find was printed at Luoyang, as if the matter had now been resolved.Footnote ³¹

The differences of opinion were particularly marked at a conference held in 1999 at Yonsei University in Seoul on the origins of printing in East Asia.Footnote ³² Much of the debate between Chinese and Korean scholars focused on the paper on which M was printed. Korean paper was sent to Chang'an in tribute by Silla, so even if the paper could be shown to be of Korean manufacture, the sūtra could still have been printed in China.Footnote ³³ What is striking about these and other arguments advanced in 1999 and subsequently is that they are all predicated on the assumption that the Pulguksa sūtra was printed in the eighth century. Silently, then, its antiquity has been accepted, and the dispute has focused on its origins instead.Footnote ³⁴

In 2007, however, new evidence came to light which showed incontrovertibly that the Sŏkkat'ap pagoda at the Pulguksa had not been left untouched since 751 but had been opened and repaired in the early eleventh century. The mass media in Korea took this discovery as evidence that the Pulguksa sūtra had been inserted into the pagoda when the repairs were made. Kim Sŏngsu has vigorously rebutted these suggestions, demonstrating that it was precisely in the eighth and ninth centuries, and not later, that M can be shown to have inspired the production of pagodas in Korea, in some cases with a copy of M placed inside.Footnote ³⁵ The balance of probabilities remains that it was printed in the eighth century, not the eleventh.

There are no grounds for assuming that the scroll could not have been printed in Korea. On the contrary, it is arguable that, if printing on the scale of the HD was possible in eighth-century Japan, it was a fortiori possible in the technologically more advanced state of Silla. Furthermore, since it appears that the Pulguksa sūtra contains a number of errors and omissions, this argues more plausibly for production in Kyŏngju than in China, where the sūtra was translated.Footnote ³⁶

From the perspective of the history of printing in East Asia, however, this is all in some sense a sideshow, for Japanese and Korean scholars agree that the technology did not originate in the periphery but came from Tang China. From this point of view, the significance of both the HD and the Pulguksa sūtra lies in the light they cast upon the development of printing in China, since no physical evidence in the form of printed artifacts that are of indisputably Chinese origin and can be accurately dated has so far come to light. It has been claimed that a printed Sanskrit dhāraṇī found in Chengdu, Sichuan, in 1944 dates from the eighth century, but this dating has been challenged.Footnote ³⁷ Other printed dhāraṇī, both in Sanskrit and in Chinese, have come to light since then, but these, too, lack any evidence relating to their dates.Footnote ³⁸ Pan Jixing has drawn attention to one found in a tomb thought to date from the seventh century and to other examples that may have been printed in the seventh century, but again there is no certainty to be had.Footnote ³⁹

In recent years Welch, Barrett, Chen, and Strickmann have teased out references to printing in early Buddhist and Daoist texts.Footnote ⁴⁰ In particular, the first three have turned the spotlight upon empress Wu, during whose reign M was translated, and upon Fazang 法藏 (643–712), who was the third patriarch of the Huayan 華嚴 school. Fazang was also the co-translator of M, and unmistakable metaphors involving printing have been identified in his writings. Although this painstaking work has brought much valuable data to light and has presented us with a persuasive hypothesis from Barrett and Chen concerning the connections between Wu, Fazang, M and printing, it remains a hypothesis and we must await the emergence of further evidence. Nevertheless, if Chinese printing technology was transmitted to Japan by the mid-eighth century and put to use for the HD, it must surely have been by Wu's time, at the latest, that the technology was developed and first put to use in China.

Both the Pulguksa sūtra and the HD involve the same recent Chinese translation of the same sūtra. This is unlikely to be a coincidence, and strongly suggests that this sūtra had already been printed in China. Indeed, in a remote library in Japan there is what is said to be a Tang printed copy of one of the spells contained in M; when it was brought to Japan is unknown, but it is evidence of the printing of M in China.Footnote ⁴¹ In light of all this, the very nature of M seems to be of critical importance, particularly since it was translated into Chinese during Wu's reign, so it is to M that I shall now turn.

THE DHĀRAṆĪ SŪTRA AND ITS BACKGROUND

The Sanskrit original of M does not seem to be extant, but several Tibetan translations do survive, and from these the original Sanskrit title has been tentatively reconstructed as Raśmivimalaviśuddhaprabhā-dhāraṇī sūtra.Footnote ⁴² It was recorded in 722 that it had been translated into Chinese by a Tokharian monk called Mitraśānta 彌陀山, together with Fazang, in the “last years” of the reign of empress Wu.Footnote ⁴³ In the form of this Chinese translation, M is known to have reached Silla very rapidly, for an inscription in a reliquary found in 1942 states explicitly that a copy was placed inside a pagoda at the Hwangboksa 皇福寺 temple in Kyŏngju in the year 706; we do not know whether this was a manuscript or a printed copy, for the original does not survive.Footnote ⁴⁴ It is not known when M first reached Japan: it is mentioned in documents preserved in the Shōsōin dated 737.Footnote ⁴⁵ Of course, it might well have arrived earlier: according to a record preserved in a later collection, four miniature pagodas each containing a dhāraṇī from M were placed inside a pagoda built at the behest of Shōtoku's mother, Kōmyō 光明, in 730.Footnote ⁴⁶

The impact of M on Buddhist practice in East Asia lasted well beyond the time of its translation, to judge by the few references available to us. Shi Siming 史思明, who was the President of the Censorate before he became one of the leaders of the An Lushan rebellion in 755, was ordered to show reverence to a pagoda designated with the title of M, in aid of peace in the land.Footnote ⁴⁷ Somewhat later, in 799, a similar pagoda was built under the inspiration of M and in the Khitan empire (Liao dynasty) during the eleventh century M was used for the names of several pagodas.Footnote ⁴⁸

What is it about M that gave rise to these devotional practices, to the production of pagodas and to the association with printing? M is a short text in which the Buddha recites six dhāraṇī.Footnote ⁴⁹ The first, the “basic” 根本 spell, is to be copied out 77 times, the copies being placed either in a large pagoda or in 77 small clay pagodas: “people who are about to die will prolong their lives to old age, all their previous sins and evil deeds being completely destroyed”. The second, 相輪, is to be copied out 99 times and the copies placed inside small clay pagodas to achieve “a great accumulation of virtues and blessings”; the third, 修造佛塔, is for recitation and copying when repairing pagodas; the fourth, 自心印, is to be copied out 99 times or recited up to 11,000 times; the fifth, which is unnamed, is merely to be recited in order to prolong one's life, purify one's body of filth, dispel illness and avert disasters; the sixth, also unnamed, is for reciting or copying out in unspecified quantities. Further, it is suggested that “anyone wishing to complete the six ways to salvation … should copy out ninety-nine copies of the first four above mentioned [dhāraṇī], make 99 small pagodas, and insert a copy in each”.Footnote ⁵⁰

The third and the fifth of these dhāraṇī are not included in the extant HD spells. What is more, M makes no reference to printing, paper, or wooden pagodas, let along the figure of one million. The rituals it prescribes, however, have their origins in practices that were once widespread throughout the Buddhist world and were transmitted eastwards to Korea and Japan.

The best-known account of the production of large quantities of stūpas or pagodas is the story of Aśoka's distribution of 84,000 stūpas. Although this story was well known in Japan, at least through the translation of texts recounting the Aśoka legend, it does not seem to have cast much of a shadow, at least until the twelfth century.Footnote ⁵¹ Therefore, in the search for the antecedents of the HD it is necessary to cast our net a little wider and consider popular practices involving stūpas and dhāraṇī. Our documentary knowledge of these practices in India is largely dependent on references in a number of sūtras and on accounts left by travelling Chinese monks, although they are in fact supported by archaeological finds.

As is well known, Xuanzang 玄奘, who visited India in the seventh century, recorded in Da Tang xiyu ji 大唐西域記 as follows:

Similarly, Yi Jing 義淨, who traveled at the end of the seventh century, wrote:

There can be no doubt that behind these accounts lay some knowledge of actual practices in India, for miniature clay stūpas with Buddhist texts inside have survived from sites in India, as also have impressions produced in clay by seals or stamps containing some form of Buddhist text, including in some cases dhāraṇī. The oldest extant examples have been tentatively dated to the first century bce, but they appear to have been produced up to the tenth century. In one sense we are fortunate to have any at all, for “it was the act of manufacture which mattered most. What was done with the completed plaque [clay tablet] was of secondary importance”.Footnote ⁵⁵ It is here that the roots of the HD lie. After all, these clay tablets with impressions of a seal or stamp are evidence of the use of mechanical means for the mass production of texts which were not manufactured for the purpose of reading; once they had been produced and ritually used their fate was of little concern. Clay was not the only material used, however, for “from a very early period are found votive offerings on gold or silver inscribed with the Buddhist creed, which would appear to have been placed in stūpas or buried in the foundations of monasteries or similar religious foundations”.Footnote ⁵⁶

Miniature stūpas have been found not only in India but also in Afghanistan, South-east Asia, Central Asia, and Tibet, sometimes with impressions of dhāraṇī in clay placed inside.Footnote ⁵⁷ Stein found similar examples in both wood and clay in Khadalik, Sorchuk and other sites, and recorded encountering “a mass of miniature stūpas in clay between two and three inches in height and all uniform in shape”.Footnote ⁵⁸ And the Turfan collection in Berlin contains a number of miniature wooden stūpas from various sites on the Northern Silk Route, of which the earliest are tentatively dated to the sixth century.Footnote ⁵⁹

These practices have their origins in the relic cult in Indian Buddhism, and as the stūpa came to symbolize the Buddha and Buddhist teachings, bodily relics placed inside stūpas came to be replaced by physical examples of those teachings, at least by the early part of the first millennium. A growing body of Mahāyāna texts expounded the various merits of these practices, which in time spread to Tibet, China, Korea and Japan.Footnote ⁶⁰ This, then, was the meaning of manufacturing miniature stūpas and placing inside them texts that nobody was supposed to read. In East Asia these practices underwent several changes, the most obvious of which was the transformation of the stūpa into the pagoda.Footnote ⁶¹ What concerns us next, therefore, is the transmission of these ideas and practices to East Asia, and ultimately to Japan.

As Barrett has pointed out, there are difficulties in taking Yi Jing's testimony as an eyewitness account.Footnote ⁶² The reference to paper is an obvious problem, for although papermaking may have been practised in India as early as the sixth century, it must be supposed that production was localized, for the necessary raw materials were only to be found in the Himalayas, which Yi Jing did not visit.Footnote ⁶³ While it remains possible that he was depending on reports he had heard while in India, far more persuasive is Barrett's argument that his references derive from Chinese sources and are addressing Chinese agendas.Footnote ⁶⁴ Nevertheless, this would not have been obvious to a Chinese reader of the text who was removed in time or space, let alone to a Japanese reader in the eighth century, who might well have concluded from this text that paper was an acceptable material to use for devotional purposes. Owing to the suppression of Buddhism at the end of the Tang, little trace has yet come to light in China of devotional practices involving miniature stūpas. However, from Yunnan some spells, including parts of M, inscribed in Sanskrit on bricks have been found inside pagodas, and have been assigned by Liebenthal to some time between 800 and 1100; although postdating the HD, they do at least testify to the practice even in remoter regions of China.Footnote ⁶⁵

The merits of constructing miniature pagodas were, in fact, well known to East Asian Buddhists through the translation of other texts expounding those merits, such as the Zaota gongde jing 造塔功徳經. This was translated into Chinese by Divākara 地婆訶羅 (613–688), who reached Chang'an in 680 and some of whose many translations carried prefaces by empress Wu before she began ruling in her own right. We can be certain that this text had not only reached Japan but also been much copied before 770, for it crops up repeatedly in records preserved in the Shōsōin.Footnote ⁶⁶

More pertinent in connection with the HD is the Lakṣacaityasamntpatti, or “One hundred thousand caitya observance”. This was translated into Chinese between 695 and 704, in other words during the reign of empress Wu, under the title Baiqianyin tuoluoni jing 百千印陀羅尼經 by Śikṣānanda 實叉難陀, a Khotanese born in 652 who was resident for some time in Luoyang.Footnote ⁶⁷ This text explains how to maintain and demonstrate one's faith in times of unbelief and expounds the benefits of multiple production up to one million caitya/stūpas and the possible use of wood, precious metals and other materials to produce them.Footnote ⁶⁸ This is the only text known to me which mentions the figure of one million and the possible use of wood, two features of the HD, and its presence in Japan in the Nara period is again clear from its mention in documents preserved in the Shōsōin.Footnote ⁶⁹

In addition to these texts which are known to have been transmitted to Japan, there is also the possibility that the practices they describe might have been transmitted to Japan by human agency. There was, after all, a considerable traffic in scholars and clerics between Japan and the mainland in the eighth century, and we know that texts and ritual objects crossed the seas with them to Japan; doubtless knowledge of other practices was passed on orally or by example, but that remains out of reach.Footnote ⁷⁰ The most important, perhaps, are the Buddhist immigrants and especially the only Indian known to have reached Japan by the eighth century, the man known in Japanese as Bodaisenna 菩提僊那 (704–760). He reached Japan in 736 and in 751 he was appointed sōjō, the highest position in the state-regulated Buddhist hierarchy, and as such presided over the inauguration of the Great Buddha at the Tōdaiji. In 758, at the ceremonies marking Shōtoku's abdication at the end of her first reign, he proposed that honorific titles be bestowed on both her and her mother, Kōmyō.Footnote ⁷¹ Given his high position it seems unlikely that he and Shōtoku would have been strangers, and in this connection it is worth noting that not only could he write Chinese, which thus functioned as a means of communication, but he was also said to be “highly accomplished with ju 呪 [spells]”. Although he died some years before the HD were initiated, he must be considered one of the potentially most important sources in Japan of knowledge of Indian Buddhist practices.Footnote ⁷²

What of the text of M itself? Extant records show that there were at least three copies of M in Nara in the eighth century: one in the Dairi 内裏 (the monarch's private space) itself, one in the Shibi Chūdai 紫微中台 (an office created for Shōtoku's mother, Kōmyō, in 749), and one in the possession of Genbō, who spent eighteen years in China, was politically active in the palace on his return and died in 746.Footnote ⁷³ In the absence of any indication to the contrary, we must assume that these were manuscripts, but it is far from inconceivable that they were copies printed in China or Silla. Their presence at the heart of the political machine in Nara is suggestive. All the same, it is clear that in many respects the HD did not follow the practices recommended in M, so the HD cannot be seen simply as a ritual act following M.

ONE MILLION MINIATURE PAGODAS?

How credible is the claim in the Shoku nihongi that one million pagodas and spells were produced? Only the Hōryūji, of all the temples to which they are said to have been distributed, still possesses pagodas and spells, but they fall well short of ten thousand, let alone one million. And where did the figure of one million come from? Most attempts to explain it have smacked of desperation. Miyasaka, for example, claims that M recommends 99,000, then assumes that the larger wooden pagodas were used to mark each lot of one hundred thus producing a total of 100,000, and multiplies this by ten for each of ten great temples.Footnote ⁷⁴ The Tōdaiji yōroku tells us that the pagodas and spells were distributed to each of “ten great temples”: what precisely were these temples and is there any evidence that they ever contained a share of the HD?

The expression Ten Great Temples, which occurs in the two accounts of the HD in the Tōdaiji yōroku, is at once more explicit than the vague reference to “various temples” in the Shoku nihongi and frustratingly inexact. The term Great Temple was not one that necessarily reflected the size or magnificence of a given temple; rather, it had political overtones, connoting a temple enjoying state patronage, or at least connections. This was true in Tang China, where the connotation of the equivalent term da si was similar, and it is without doubt a reflection of Tang practice that this term came into use in Japan.Footnote ⁷⁵ Further, the growing number of Great Temples in the Nara period is most probably to be taken as an emblem of the state's, or at least the court's, patronage of Buddhism. Given that the HD were promoted by Shōtoku, it is most likely that the pagodas and their contents would have been presented to state-patronized temples, in other words, in contemporary parlance, Great Temples. We must, therefore, register the possibility that the compilers of the Shoku nihongi replaced 大寺 with 諸寺 deliberately, to disconnect the HD project from the Great Temples and thus from the court and the state.Footnote ⁷⁶ Similarly, we must register the possibility that the absence of terms such as Ten Great Temples from the Shoku nihongi as a whole indicates an attempt to play down the connections between Shōtoku's court and the Buddhist establishment in the mid-eighth century.

The HD were not regarded as valued treasures in later years, for they disappear from recorded history even in the case of Hōryūji. Although we cannot draw any firm conclusions about the Ten Great Temples and their putative consignments of the HD pagodas, we can at least narrow down the range of possibilities. The minimalist position might be that, whatever the initial objectives might have been and whatever the Shoku nihongi says, only the Hōryūji's batch of 100,000 was ever produced; this is the line taken by Nagasawa.Footnote ⁷⁷ This seems untenable, for unrelated sources state, without making much of the fact, that 100,000 miniature pagodas (nothing is said of the spells they may have contained) were at one time housed in the Tōdaiji and the Yakushiji, that 84,000 were once housed in the Gangōji, that 100,000 were at one time housed in the Kōfukuji and 20,000 in the Shitennōji, and that there was a Small Pagoda Pavilion in the Saidaiji, which in all likelihood held some of the HD.Footnote ⁷⁸ The maximalist position, that a full one million were produced and distributed to the ten temples in batches of 100,000, runs up against the silence of the sources on the three remaining temples of the ten, the Gufukuji, Sūfukuji and Daianji. Given that the Hōryūji would be an identical case were it not for the survival of the HD themselves, the silence should not be regarded as conclusive negative evidence. On the evidence presently available, all that we can say is that a conclusion closer to the maximalist position than the minimalist seems justifiable. Thus whether precisely a million were made or not, we are justified in supposing that a million was the goal and that in all likelihood several hundreds of thousands at least were made.

All the records we have speak only of the miniature pagodas; it is assumed that all the pagodas, like the ones in the Hōryūji, contained printed spells; yet none of our sources bothers to mention them. Still, such large quantities of wooden pagodas are impressive enough: at least 500 would have had to be made each day in order to produce one million by 770, if production commenced in 764. It is obvious from the marks on their bases that the pagodas were turned on lathes, and more than 100 different lathes can be identified from these marks; this in itself is no surprise, for lathes had been in use in Japan for several centuries by this time.Footnote ⁷⁹ What should enter our calculations now is the sheer quantities of wood needed to produce them and the area of ground they would have covered when completed. To manufacture even a hundred thousand would have required approximately 1,500 cubic metres of timber and, when placed on the ground, they would have covered an area of some three thousand square metres.Footnote ⁸⁰ It is evident, then, that this must have been a state enterprise on a huge scale, involving the transportation of very considerable quantities of timber, and at vast expense. It will be recalled that the Shoku nihongi records that 157 officials and artisans were rewarded, but we know that far more people were involved merely from the names of the artisans recorded on the bases of the extant pagodas.Footnote ⁸¹

Given the scale of the whole undertaking, the Shoku nihongi is surprisingly reticent on the HD, to a degree that suggests political interference with the historical record during the final process of compilation in 694–697. It is not only the omission of any reference to the new technology of printing or to the Ten Great Temples of state, but also the lack of any mention of the initiation of the HD project in 764, or of the subsequent stages leading to its completion. The Shoku nihongi is replete with edicts and statements of intent, so why are the HD only mentioned at their completion in 770? After all, even the brief account in the Tōdaiji yōroku, quoted above, provides a date for the commencement of the project and for the construction of two buildings to house the Tōdaiji's share of the pagodas. The Shoku nihongi's silence suggests some tampering with the record, and there is, in fact, some evidence to support this suggestion.Footnote ⁸² In the first place, the compilers, Fujiwara no Tsugutada 藤原継縄 and Sugano no Mamichi 菅野真道, stated that the part dealing with the reign of Shōtoku had been completed before they began work but that, finding it unsatisfactory for unspecified reasons, they had revised it in 794.Footnote ⁸³ Second, the reigns of Shōtoku and her successor, Kōnin, as presented in the Shoku nihongi are unusual in that both are followed by an appraisal of those reigns, in Shōtoku's case critical of a monk named Dōkyō 道鏡 (c. 700–772) and of the impact of Buddhism on the exchequer, but in Kōnin's case favourable.Footnote ⁸⁴ Unvarnished the record is certainly not.

What is indisputable is that the HD were forgotten, and found no place in subsequent histories of Japan. The HD first resurface in documentary history more than nine hundred years later, in 1717 and in public in 1842, when some of the HD pagodas were included in a public exhibition in Edo.Footnote ⁸⁵ While the dhāraṇī are now celebrated by historians of printing, the pagodas are often forgotten, but in terms of costs and resources they must have consumed the lion's share. It is time, therefore, to consider why Shōtoku should have undertaken this costly project.Footnote ⁸⁶

SHŌTOKU

The Nara period was remarkable for a succession of women rulers who occupied the throne for twenty-nine of its seventy-four years. What distinguishes Shōtoku from the others, however, is that in 738 she was named heir to the throne at the age of twenty-one, the only woman so designated. However, some have argued that she was a figurehead while power lay in the hands of Dōkyō. Shōtoku's resumption of the throne and Dōkyō's rapid fall from grace following her death in 770 cast doubt upon this interpretation. The prevailing assumption is that Shōtoku and Dōkyō were lovers. This has its origins in the Nihon ryōiki, compiled by the monk Kyōkai 景戒 in the early Heian period. Kyōkai states that Dōkyō “had intercourse with the empress on the same pillow”, and concludes that lewd songs circulating at the time “were a prediction of his relations with the empress and his control over state affairs”.Footnote ⁸⁷ This is feeble evidence and the supposed sexual relationship between Dōkyō and Shōtoku is not open to a definitive answer, but it has been common to connect Shōtoku's career to those of Fujiwara no Nakamaro and Dōkyō instead of examining it independently. There are very good grounds for rethinking Shōtoku's reign not as that of a pawn but as that of a woman quite capable of making her own political choices, including the decision to undertake the HD project, and as that of a monarch needing to position herself in an age in which religious and political issues were fraught and in which there were few models of female sovereignty.Footnote ⁸⁸

Shōtoku's motive for undertaking the HD is often associated with the “remorse theory” mentioned in the Tōdaiji yōroku, but there are in fact several problems with this theory.Footnote ⁸⁹ First, M and the practices it encourages are not connected with a public expression of remorse but with prolongation of life. Second, why should the particular case of the suppression of the Nakamaro rebellion have occasioned such a massive act of repentance when the many other instances of state violence during these years did not? And third, since M explicitly recommends much smaller quantities of pagodas to achieve the desired effect, can it really be said that the enormous scale of the HD project was commensurate with the notion that this was merely an act of remorse? The reference in the Tōdaiji yōroku to an oral tradition associating the HD with remorse thus seems misleading; any religious motivation is more likely to be connected with the health of Shōtoku in the last years of her life and the focus of M on longevity.

The eighth century in Japan was a period of continuous political strife and a relentless succession of assassinations, executions, forced suicides, plots and accusations of sorcery. And that is to say nothing of the epidemics, famines and other disasters which often prompted religious or political responses.Footnote ⁹⁰ Shōtoku was no caretaker on the throne, but, as the political strife and the documentary record show, her tenure, particularly after her resumption of the throne in 764, was undoubtedly precarious. Moreover, she had to deal with a political apparatus that was entirely male and it is difficult to suppose that her gender was irrelevant and that she did not stand in particular need of allies and symbols. Two allies can be clearly identified: one was Kibi no Makibi 吉備真備 (695–775), who had spent the years 717–735 in China as a government student. He served as Shōtoku's tutor for a while and rose under her patronage. In 764 he was appointed chief of the bureau for the construction of the Tōdaiji and was responsible for crushing Fujiwara no Nakamaro's rebellion. In 766, he was appointed minister of the right, “in recompense for teaching us since we were heir apparent”, and his daughter tended Shōtoku in her final illness.Footnote ⁹¹

Another ally was of course Dōkyō. He is a figure of some interest, in spite of the opprobrium that has been heaped on his head for centuries. He was born at the beginning of the eighth century, and there is an ancient tradition that he was a grandson of Tenji (r. 661–672).Footnote ⁹² According to his unflattering biography in Shoku nihongi, he had a knowledge of Sanskrit; he entered Shōtoku's service in 761 and rose rapidly, attaining the pre-eminent and unprecedented positions of daijō daijin zenshi 太政大臣禅師 in 765 and hōō 法王 in 766.Footnote ⁹³ Horiike attributes the HD project to Dōkyō.Footnote ⁹⁴ There can be no doubt of Dōkyō's credentials and it is certainly the case that Dōkyō was familiar with M, for there survives in the Shōsōin a request he made for a copy of M which is dated 763.5.16.Footnote ⁹⁵ This is just one year before the HD project is supposed to have begun. Impressive though this is, it needs to be remembered that in terms of print, paper and quantities, M was clearly not the sole inspiration: the huge quantities and the use of wood might have been suggested by the Lakṣacaityasamntpatti which had been translated into Chinese during the reign of empress Wu and was certainly available in Japan during Shōtoku's two reigns, but where did the knowledge of printing come from? If some printed dhāraṇī had by this time been transmitted from China, perhaps by Kibi no Makibi, these would have shown Dōkyō and others that it was possible to replicate texts mechanically, but it would not have shown them how this was to be done. That, presumably, only those who had been to China (or possibly Korea) and had seen it done would have been able to explain.

This argument, however, assumes that the same technology was used in China and in Japan, and here we run up against the metal-plate theory for the production of the HD. Since the Pulguksa copy was indubitably printed with woodblocks and is now widely considered to have been printed before the HD, it is this technology that Japanese are most likely to have come across. The use of metal plates would therefore have to be considered an autochthonous development in Japan and need not have depended upon first-hand experience in China. Given that the text of M from which the spells were taken had been translated in Chang'an not seventy years earlier and that numerous Japanese had visited Chang'an over that period, the balance of probability surely lies with the transmission not merely of the text of M but also of the technology directly from Chang'an, and thus with the theory that the HD were printed with woodblocks. We may never know just how and when the technology was transmitted to Japan. Nevertheless, the scale of the project and the use of printing technology suggest that it was not so much an act of Buddhist contrition as an assertion of power and of control of resources and thus an act of symbolic significance. Needless to say, it may well have served both public political goals and private devotional goals simultaneously.