[In the following essay, Sato traces Teika's contribution to the tradition of composing long sequences of tanka (or waka) poetry, the forerunner of the later renga form.]

In the following argument, the tanka, “short song,” is a 5/7/5/7/7/syllable poetic form, which came into being during the seventh century at the latest and has remained in use ever since. Early on it became the predominant verseform for court poets. English translators usually regard the tanka as a five-line poem because it consists of five syllabic units but Japanese poets and scholars take it to be a one-line poem.

The renga, “linked song,” is the group-oriented verseform which came into being as a result of the tendency of the tanka to break up into the upper hemistich of 5/7/5/syllables and the lower hemistich of 7/7/ syllables. Initially a tanka composed by two persons, one hemistich by one hand, the other by another, the renga eventually grew into a sequential form of 100 alternating hemistiches or units (equal to 50 tanka) with extraordinarily complex rules, at times composed by a dozen poets.

By some unaccountable academic habit, Japanese scholars seem to annotate certain classical literary texts with what might be called repetitive abandon, while virtually ignoring the others. A primary example of this lopsided approach is the “eight imperial anthologies” (Hachidaishû), of which the first and the eighth are treated with an indulgence alien to the six others.¹ The “six personal collections” (Rokkashû) do not fare much better. Of the Chôshû eisô of Fujiwara no Shunzei (1114-1204), Sankashû of Saigyô (1118-1190), Shûgyokushû of Jien (1155-1225), Minishû of Fujiwara no Ietaka (1158-1237), Shûi gusô of Fujiwara no Teika (1162-1241), and Akishino gessei shû of Fujiwara no Yoshitsune (1169-1206), Saigyô's collection can be read in at least four readily available annotated editions, three of them published in the past ten years, but the five others have received no comparable attention. This has been frustrating for those students of Japanese literature who must rely on specialists' extensive annotations to understand classical poems.

Kubota Jun's first fully annotated text of the “complete poems” by Fujiwara no Teika, Yakuchû: Fujiwara no Teika zen-kashû, (Kawade Shobô Shinsha, 1985-86), which was recently published, fills a considerable gap in this regard.² The effort is also epochal. Teika is indisputably the greatest poet of the Shinkokin period,³ and Kubota, surely the most knowledgeable living scholar of the poetry of the same period.⁴ In view of the complexity and difficulty of Kubota's task—reflecting in 1983 on the work that had taken him more than a decade, he said that it was an “improbable undertaking”—⁵ and the magnitude of his achievement, any similar attempt in the coming decades is likely to be a pale copy of what he has done.

To describe Kubota's effort briefly, the verse section consists of Teika's own compilation Shûi gusô (3663 pieces) and later gleanings, and its arrangement is almost the same as that in Reizei Tameomi's unannotated edition of Teika's “complete poems,” originally published in 1940.⁶ The differences occur in the gleanings. Kubota has dropped a loose sequence of 351 tanka on hawks as unlikely to have been composed by Teika, and added 37 which were previously uncollected. Altogether, Kubota's text contains 4257 pieces; Reizei's, 4571.⁷ All of them are tanka, except one, which is a chôka.

Aside from translation and annotation (yakuchû), however, Kubota's edition has far more to offer than Reizei's. Reizei was content to provide a general discussion of Teika's poetry, its bibliographic sources, a chronological table, and an index of poems. Kubota gives, in addition, what appears to be an exhaustive listing of anthologies, poetry matches (utaawase), poetic tales, diaries, and other texts where Teika's pieces are cited. This he says he does in order to show how Teika's poetry was “appreciated and enjoyed in the medieval age.” He also provides a complete list of poetic place-names (utamakura) used in Teika's poems, with helpful comments. These two lists, on top of the abundant citations in headnotes and supplementary notes of tanka directly alluded to (honka), related tanka, and other references, make Kubota's two volumes a marvel of thorough research and erudition.

This edition of Teika's poems, with its chronological table of unusual clarity, gives us an opportunity to consider one possibility I have not seen discussed until now: that the emphasis Teika and the poets of his time placed on composing tanka in sequential formats, rather than individually, and doing so in competitive settings, may have played a pivotal role in the creation of the renga form.⁸ Because of Teika's detailed and copious diary, Meigetsuki, it has often been pointed out that renga developed in a few decades during Gotoba's reign from a brief composition of a few links to lengthy “chains of links.”⁹ The diary also shows that during the latter part of his life Teika developed a passion for composing renga with his friends, while his interest in tanka waned and became more or less perfunctory. But, to my knowledge, a possible creative transition from tanka composed as sets of substantial lengths to the renga form has not been suggested.

Quite early on the brevity of the tanka form must have spawned the urge to make each piece a part of a greater versifying whole. Groups of tanka composed on certain topics and more or less sequential “diary pieces” scattered throughout the Man'yôshû—the eighth-century anthology of more than 4500 poems—are perhaps the earliest manifestations of this urge. Weaving tanka into episodic prose narratives, as in Ise monogatari (9th century), is another such manifestation. By the second half of the tenth century such poets as Minamoto no Shitagô (911-983), Sone no Yoshitada (fl. late 10th century), and Minamoto no Shigeyuki (died around 1000) had clearly shown their dissatisfaction with the 31 syllables as an independent poetic vehicle by devising some ambitious sequential formats. Shitagô wrote, for example, a 48-piece set which used the ame tsuchi syllabary¹⁰ to form both an acrostic and a telestich.¹¹ According to his own headnote, the set was composed in response to a similar one which the poet Fujiwara no Aritada (dates unknown) sent to him. He proudly added, however, that Aritada's set incorporated the ame tsuchi syllabary only to form an acrostic, whereas his own was made to form a telestich as well. Moreover, he said he divided the 48 pieces into seasons (toki)—in fact the four seasons plus reflections (omoi) and love, a total of six categories.¹² Yoshitada, on the other hand, wrote the longest diary piece known from that period, a cycle of 360 tanka. He did this by dividing each month into three equal sections of ten days and writing ten tanka for each section, or one tanka a day. He introduced each season with a chôka and an envoy in tanka form.¹³

The three poets mentioned also wrote what was to become the standard, formal unit of both tanka and renga composition: a 100-piece set called hyakushu. Which of the three wrote a proper hyakushu-uta first remains in doubt, with some asserting that “there should be no doubt” that it was Yoshitada.¹⁴ However, the following observations can be made. Yoshitada's set is somewhat loose in structure as it consists of 10 tanka each on spring, summer, autumn, winter, and love, plus 31 tanka incorporating a tanka to form an acrostic and another to form a telestich, 10 tanka on jikkan (ten calendar signs) and 10 more on directions. Thus it not only lacks a sense of proper organization, but it actually consists of 101 tanka. The set Shitagô wrote in response, which is included in Yoshitada's collection in its entirety, follows exactly the same format. In contrast, Shigeyuki's set consists of 20 tanka on each of the four seasons, and 10 each on the two subjects of love and “resentment” (urami).¹⁵ Moreover, neither Yoshitada's nor Shitagô's is called a 100-piece set, but Shigeyuki's is.

About the same time or not long afterward, Shigeyuki's Daughter (dates uncertain), Izumi Shikibu (969-?), and Sagami (998-1056?), among others, wrote 100-piece sets.¹⁶ Of the 100-piece sets left by these female poets, the three by Sagami may be the most notable as they are written to express a single state of mind: a woman's anxieties and concerns. Izumi tried sets of various kinds and lengths, among them a sequence of 45 tanka describing her longings for a dead lover—from noon, evening, night, midnight, to daybreak—which was evidently written or assembled to give a sense of temporal continuity.¹⁷

The 100-piece set as a format for tanka composition acquired official sanction with the Horikawa hyakushu, a group of 14 sets of 100 tanka composed from 1105 to 1106 at the request of the emperor Horikawa (1079-1107). This set became a monument in the history of tanka because it not only gave categorical breakdowns (spring, 20 pieces; summer, 15; autumn, 20; winter, 15; love, 10; and miscellaneous topics, 20) but also specified a topic for each of the hundred pieces. Within decades, other formats came into being, such as the Eikyû hyakushu in 1116 (spring, 18 pieces; summer, 12; autumn, 18; winter, 12; love, 10; and miscellaneous, 30, with all the topics specified), and the Kyûan hyakushu in 1150 (spring, 20 pieces; summer, 10; autumn, 20; winter, 10; love, 20; and miscellaneous, 20; no topics specified). But it was the Horikawa set that became a model for later poets, as well as a format to be used for practice or demonstrating poetic skill. Teika's father, Shunzei, for example, used this format in his first known attempt to write a 100-piece set (1136-37), for his first 100-piece set that survives in its entirety (1139-40), and for five sets on five famous shrines (1190).¹⁸

So, composing tanka in certain formats or to create certain patterns, such as acrostics, had had two centuries of history by the time Teika began to write poetry. Yet it was about this time that these “format compositions” became routine and, for a while, even a vogue.¹⁹ This is partially reflected in the Rokkashû: in the collections of the two poets in the earlier generation, Shunzei and Saigyô, format compositions do not predominate; in those of the four others, they do.

The 480 tanka in Shunzei's Chôshû eisô, first put together in 1178, include two sets of 100 poems, Kyûan and Horikawa (the Kyûan set actually consists of 101 poems and has a chôka), and a few sets composed in shorter units.²⁰ The collection also has several references to compositions in units of ten. But the proportion of the format pieces is less than half of the total. Among the 1552 tanka in Saigyô's Sankashû there is only one 100-piece set. The group of “110 tanka on love” may not have been originally intended as a sequence.²¹ In contrast, more than three-quarters of the 6,000 pieces in Jien's collection are made up of format compositions, mostly 100-piece sets.²² About two-thirds of the more than 3,000 pieces by Ietaka are similar format pieces.²³ Over three-quarters of the 3663 pieces in Teika's own collection²⁴ are poems composed to accommodate certain formats. And such compositions take up about 70٪ of the 1,600 tanka in Yoshitsune's anthology.²⁵

Needless to say, in considering such anthological make-ups, we must keep in mind that the poems left to us in personal collections, official anthologies, records of poetry matches, and others, may represent only a small portion of those actually composed. In his edition of Teika's poems, for example, Reizei says Teika, a professional poet, probably composed “tens of thousands of tanka” in his lifetime but that most were lost in large part because impromptu tanka composed in response to other poets' pieces were often not kept in duplicates.²⁶ Still, it is evident that in Teika's time format compositions, which were until then represented mainly by the two 100-piece formats of Horikawa and Kyûan, blossomed in great variety with remarkable rapidity and went on to become more than “official” or ceremonial. In the process, longer formats themselves may have acquired a reason for being and thus generated an environment for the birth of the truly sequential form of renga.

Let us look at some of the more notable format compositions of Teika, along with related activities of his fellow poets.

Teika wrote his first known 100-piece set (nos. 1-100 in Kubota's text; same below), appropriately called Shogaku hyakushu, “beginner's 100-piece set,” in 1181, when he was 19. It was modeled after the Kyûan hyakushu. The following year he composed a Horikawa set on his father's “strict order” (nos. 3462-3561). The result was not good enough to include in the official part of his collection, as Teika noted in editing his book. But in so noting, he recalled how his parents “shed tears of gratitude” upon reading the set, Fujiwara no Takanobu (1142-1205), Jakuren (? - 1202), and others “poured words of praise,” the minister of the right Fujiwara no Kanezane (1149-1207) “wrote a letter praising it,” and Shun'e (dates uncertain) “came and shed tears, greatly touched.” Yet, he added bitterly, in the oft-quoted note, “someone's good reputation seems to last for a mere three to four years,” for soon people were accusing him of writing “fashionably new, rootless, Dharma poems.” “Dharma poems,” Daruma uta, is understood to mean poetry as inexplicable as Zen tales.

In 1186, at Saigyô's request, Teika, along with several other poets, wrote another 100-piece set, his third (nos. 201-300), from which four would be taken for inclusion in the Shinkokinshû. One of them (no. 135) was also to become surely the most famous piece in the Teika canon:

Miwataseba hana mo momiji mo nakarikeri ura no tomaya no aki no yûgure

As I look out, there are neither blossoms nor crimson leaves: by a cove, a thatched hut, this autumn evening²⁷

The following year he wrote two 100-piece sets, the second of which (nos. 301-400) was his first composition written in response to a set by someone else—this time, Ietaka—or to start a competition. During the same year Jien, who had already written several 100-piece sets, also wrote two more and attached to one of them a note saying “composed within six hours.” Composing tanka fast, or sokuei, would become a hallmark of Jien and of this period. Two years later, he sent a Horikawa set to Teika, who responded in kind (nos. 401-500). To his own set, Jien added a lengthy note saying that though the composition took him from the late afternoon of the 11th to noon of the 13th of the twelfth month, this was because he had to do it only when he could find the time (himahima ni). The two men did another round soon afterward.

Apparently inspired by Jien, several poets wrote 100-piece sets in 1190. In the third month Shunzei started the aforementioned cycle of five 100-piece sets on five shrines, which he would complete by the eleventh month. In the fourth month, Jien composed a set within four hours. It consisted of ten poems each on flowers, cuckoo, moon, snow, love, pine, bamboo, mountain hut, sea-lane, and reflections. In the fifth month, he did another Horikawa set, this time more leisurely, spending 16 days on it.

In the sixth month, on the 25th, when he had to stay home “because of some unclean occurrence,” Teika composed in six hours a set incorporating twenty words, such as asakasumi (morning haze), mume no hana (plum flowers), hatsuyuki (first snow), and Ono no sumikama (charcoal kilns in Ono), which formed an acrostic sequence of 20 pieces on spring, 15 on summer, 20 on autumn, 15 on winter, 15 on love, and 15 on miscellaneous topics (nos. 2792-2891). Teika then asked Jien and three other poets to follow suit. Jien promptly did, but by adding another technical difficulty: he incorporated, though not always in its entirety, each of the words selected to form acrostics in the tanka forming the acrostics. For example, the five pieces in the section for asakasumi would not only form an acrostic sequence, but each of the five would include either asakasumi or at least kasumi. Moreover, he spent only three hours to compose the set, in contrast to the six hours Teika had to spend, as he proudly noted at the end of his composition. The next day Teika composed in ten hours another 100-piece set, this time incorporating specified 5- and 7-syllable phrases in specified spots to make a sequence of 30 pieces on spring, 20 on summer, 30 on autumn, and 20 on winter (nos. 2892-2991). Again he asked Jien and others to follow his example. Again Jien did his much faster, spending only two hours; and this time, too, he noted with pride the discrepancies in talent between himself and Teika.²⁸ Jien was surprisingly competitive for a man of the cloth who would soon head the Tendai sect, later attain the rank of archbishop (daisôjô), and, bruised by his emperor friend Gotoba's military venture, failure, and exile, end up writing a reflective history of Japan, Gukanshô.

This year saw two other occasions for composing 100-piece sets, both hosted by Yoshitsune. In the ninth month, Teika and several other poets wrote what would be called Kagetsu hyakushu, 50 pieces on flowers, 50 on the moon (Teika, nos. 601-700). In the twelfth month, he and Yoshitsune each wrote in two nights a set, 5 pieces on each of 20 topics; in Yoshitsune's case, 60 pieces on the night of the 15th, 40 on the night of the 19th. Teika's set from this occasion was lost, but from this year on his poetic association with Yoshitsune would deepen.

In the following year, for example, in the second month, he and Yoshitsune exchanged memorial poems on the first anniversary of Saigyô's death (no. 3893). Though on that occasion each wrote just one tanka, in the sixth month, “on the night when the moon was bright,” Teika received from Yoshitsune an acrostic sequence of 47 tanka incorporating the i-ro-ha syllabary²⁹ with a request that he respond in kind; he did, probably on the spot, for he appears to have kept Yoshitsune's messenger waiting (no. 2992-3038).³⁰ He composed another set the next day (nos. 3039-85) because Ietaka, on reading the sequence he submitted to Yoshitsune, composed his own and sent it to him. Not long afterward, in the same month, he responded to another request from Yoshitsune and sent a set of 20 poems.³¹ And for the third time in the same month he composed a set of 15 poems, each incorporating a specific character or its meaning, such as tree, fire, or earth (nos. 3170-84), in response to Yoshitsune's similar set. A set of 13 poems listed immediately after the 15-piece set in Teika's anthology (nos. 3185-97) may also have been made at about the same time, though the year or the month of composition is not indicated. It is another acrostic sequence, this time weaving into itself the Buddhist recitation, Na-mo-me-u, and, because it was autumn, all the poems are of that season. Teika's headnote says he composed it because he heard prayers being offered at night while visiting Yoshitsune.

Such versifying facility obviously was nothing new. It is easy to imagine that talented poets such as Shitagô, who was mentioned earlier, would not have had to spend much time to turn out more complicated sequences. But composing tanka fast was evidently not the norm in the pre-Shinkokin periods. With the Kyûan hyakushu, for example, the emperor Sutoku (1119-64) made clear in 1143 his plan to have 14 poets, including Shunzei, submit 100-piece sets, but the final submissions were not made until 1150; meanwhile three poets died and had to be replaced. As late as 1178, when about 20 poets took part in composing 100-piece sets, it was decided that ten pieces would be composed every ten days, so that it took from the third month to the sixth month to complete the cycle.³² Even during the time of Jien and Teika, the standard practice at poetry matches seems to have been to notify the participants of the topics in advance so that they would have ample time to compose a few tanka beforehand.

As the principal poets competed in turning out format compositions from the late 1180s to the early 1190s and beyond, however, one important thing is likely to have happened: the continuing search for ever more novel, perhaps more difficult, formats. From the viewpoint of technical difficulty, the use of the i-ro-ha syllabary of 47 sound units to form an acrostic sequence provides a good example. The idea, which was, at least during that period, apparently instigated by the spirited and versatile Yoshitsune, may seem relatively easy until you realize that the Japanese language at the time had few indigenous words beginning with the r-sound (ra, ri, ru, re, ro) and one basic rule in the composition of Japanese poetry (waka) was the proscription against the use of non-indigenous words. So, in using the i-ro-ha syllabary a poet had to compromise, though not to the extent of completely ignoring traditional poetics.³³ The difficulty may be discerned from the fact that only 17 of Teika's poems begin with the r-sound and all except one appear in acrostic sequences which often make the inclusion of unwanted sound units unavoidable. The one exception (no. 2761) occurs in a series of tanka he wrote to go with the Lotus Sutra he copied to commemorate the anniversary of his mother's death, but there he apparently felt safe using a word of Chinese origin. None of Jien's 6,000 tanka begins with this sound.

To see a few more format compositions from Teika's collection, the sequence of 128 poems he composed in 1196 at Yoshitsune's residence (nos. 1501-1628), has the noun (or, more precisely, the Chinese character that represents the noun) that comes at the end of each piece “rhyme”—probably with a phrase taken from an unidentified Chinese book in use at the time. The set is evenly broken into the four seasons and four other categories, a total of eight, 16 tanka allocated to each category. In 1217 Teika made a similar but different attempt: he composed a set of 64 seven-character couplets in Chinese and a set of 64 tanka to “rhyme” with them: 20 pieces on spring, 12 on summer, 20 on autumn, and 12 on winter (nos. 3398-3461).

Yet another variation of this approach is the set composed in the following year at Jien's recommendation (nos. 3198-3297): the 100 tanka in this sequence—15 pieces on spring, 10 on summer, 15 on autumn, 10 on winter, 5 each on love, mountain hut, and old village, 10 each on living alone, reflections, and transience, and 5 on Buddhism—are in effect translations of lines from Po-chu'i's poems. A set that may be cited as the last example from Shûi gusô is the 100-piece sequence Teika composed in 1220 (nos. 3298-3397): it gives four tanka to each of 25 topics selected, the four on each topic describing spring, summer, autumn, and winter. The collections of Jien and Yoshitsune contain many other kinds of formats then invented.

These sequential experiments continued for quite a long time, but dissatisfaction with tanka composition appears to have begun to be felt rather early on. The poet and scholar Yasuda Ayao noted that Teika registered his discontent with tanka composition for the first time in a poem when he wrote, in 1196, the following (no. 1583):³⁴

Koke no shita ni uzumanu na o ba nokosu tomo
                    hakana no michi ya Shikishima no uta
My name might remain unburied under the moss,
                    but such a transient way is Shikishima poetry

Shikishima is one of the old names of Japan, and Shikishima no uta a conventional name of tanka. After this the frequency of his complaints in his diary increased, often expressed in the form of utter disaffection, especially with participating in poetry meetings where he had to produce poems on demand.

A professional court poet who was highly esteemed, Teika of course continued to play his required role, attending poetry gatherings, composing poems as requested. It may be significant, however, that the poem expressing his discontent was part of the 128-piece set mentioned above, the longest and perhaps the last sequence he did with Yoshitsune.³⁵ Yoshitsune, who would eventually write the kana preface—there was also a Chinese preface—to the Shinkokinshû, was the spiritual leader of the poets of the era because of his talent, versatility, and, above all, his exalted court rank. It appears that after the autumn of 1196 his passion shifted to some other forms of poetry.

On the 18th of fifth month, 1199, Teika refers to renga for the first time in his diary. On the 22nd of the twelfth month, of the same year, he reports he did renga with Yoshitsune and another poet at Yoshitsune's, until late at night. The mentions of his participation in renga sessions suddenly increase after Yoshitsune's untimely death in 1206. Then, a few years after Gotoba's exile he begins to record his candid joy in doing renga—just as he recorded his candid disaffection with tanka in the earlier years.³⁶

The renga that Teika and his fellow poets composed are preserved only in scattered forms in the first renga anthology Tsukuba shû, compiled in 1356 by Nijô Yoshimoto (1320-88), and it is not even certain whether the standard 100-unit format was established in Teika's lifetime.³⁷ Still, it seems reasonably clear that the experiments in sequential forms of tanka he and his friends pursued opened the way to the renga form.

Notes

1. Based on the bibliographical information I have (which is far from complete), at least six annotated editions of the Kokinshû, compiled in 910, and thirteen of the Shinkokinshû, compiled in 1205, are on the market at present. In contrast, with some of the six others it is difficult to get even their unannotated texts. Though to a much lesser degree, a similar situation has obtained recently in the United States, where two complete translations of the Kokinshû have been published consecutively: one by Laurel Rasplica Rodd with Mary Catherine Henkenius (Princeton University Press, 1984) and the other by Helen Craig McCullough (Stanford University Press, 1985). No complete trans-lation of any other anthology exists.

2. Before Kubota, there were several unannotated or barely annotated texts of Teika's poems. Among them are Sasaki Nobutsuna's, in 2 vols. (Kaizôsha, 1939-40); Reizei Tameomi's (Bunmeisha, 1940; reissued in 1974, as later mentioned in the article); and Akahane Shuku's (Kazama Shobô, 1974).

3. Teika's reputation has fluctuated over the centuries, but this may be largely attributable to factional assessments. Some part of the cause may also lie in the distinction made early on between uta-yomi, a natural poet who can readily compose a poem on any topic or sentiment, and uta-tsukuri, a poet who labors to compose a poem. Saigyô was a leading uta-yomi, and Teika, a leading uta-tsukuri.

4. Among his many other books are a 1021-page study of Shinkokin poets, Shinkokin kajin no kenkyû (Tokyo Daigaku Shuppankai, 1973), and a 9-volume annotation of the Shinkokinshû (Kôdansha, 1984-86).

5. Kubota Jun, Hana no momo iu (Shinchosha, 1984), p. 140.

6. Reissued in 1974 by Kokusho Kankô Kai. The late Tameomi was Teika's descendant.

7. Without explanation Kubota has retained the 351 pieces on hawks in the numbering, so that the last piece in his edition becomes the 4608th. Both Kubota and Reizei duplicate the numberings when other poets' compositions appear, so the two numbers, 4257 and 4571, represent all Teika's poems.

8. Upon reading a draft of this article, Professor Earl Miner, of Princeton University, pointed out that this possibility has been discussed before. For example, he and Professor Robert Brower observed in Japanese Court Poetry (Stanford, 1961), p. 417: “the most common unit of poetic sequences by single authors, the hundred-poem sequence (hyakushu-uta) became the standard number of stanzas for renga practiced by the masters of the form.” As Miner was careful to note, however, my emphasis here is not so much on the numerical link between the hyakushu-uta and renga—or, for that matter, on the influence of the rules and aesthetic conventions of waka on renga writing. Rather, it is on certain activities of Teika and his poet friends which may have helped give birth to the standard renga form. Naturally, in the absence of solid evidence I can only offer my argument as a speculative possibility.

9. I have discussed this in One Hundred Frogs: From Renga to Haiku to English (John Weatherhill, 1983), pp. 12-17.

10. A collection of words meant to contain all the essential sounds in Japanese, each sound just once. Comparable to composing the shortest sentence possible containing all the alphabetical letters, each letter once.

11. I have briefly referred to this set in “Lineation of Tanka in English Translation,” Monumenta Nipponica, Vol. 42, No. 3, p. 350.

12. Shitagô's collection may be found in Nishihonganji-bon sanjûrokunin shû seisei, ed. Kusokami Noboru (Kazama Shobô, 1966).

13. Yoshitada's collection may be found in Heian Kamakura shikashû, ed. Hisamatsu Sen'ichi et al (Iwanami Shoten, 1964).

14. Hashimoto Fumio and Takizawa Sadao, Kôhon: Horikawain ontoki hyakushu-uta to sono kenkyû (Kasama Shoin, 1976), p. 334.

15. Shigeyuki's collection may be found in Kusokami's edition.

16. The collections of Shigeyuki's Daughter and Sagami may be found in Nyonin waka taikei, Vol. 2, ed. Nagasawa Mitsu (Kazama Shobô, 1960). Izumi's poems can be read in many editions.

17. No. 1014-59, in Shimizu Fumio's edition (Iwanami Shoten, 1956).

18. On the preeminence of the Horikawa set, see the book by Hashimoto and Takizawa, cited in note 13, pp. 361-62, 381-86, and 481-94.

19. The detailed chronological table provided in Fujiwara no Shunzei: Hito to sakuhin, Vol. 2, Taniyama Shigeru chosaku shû (Kadokawa Shoten, 1982) shows that between 1114, when Shunzei was born, and the late 1170s attempts at format compositions were sporadic at best.

20. Chôshû eisô in Hisamatsu's Heian Kamakura shikashû.

21. Sankashû, in Sankashû, Kinkai waka shû, ed. Kazamaki Keijirô and Kojima Yoshio (Iwanami Shoten, 1961).

22. Kôhon: Shûgyokushû, ed. Taga Munehaya (Yoshikawa Kôbun Kan, 1971).

23. Fujiwara no Ietaka shû to sono kenkyû, ed. Kubota Jun (Miyai Shoten, 1968).

24. The Shûi gusô consists of two sections, which may be described as official and unofficial. The official section has 2791 poems which Teika apparently regarded as formally acceptable pieces; the unofficial one, called ingai, 872 pieces which he thought too informal or inferior to be included in the other section, though not poor enough to be discarded.

25. Fujiwara no Yoshitsune zen-kashû to sono kenkyû, ed. Aoki Kengô (Kasama Shoin, 1976), pp. 181-98. Among Yoshitsune's 469 pieces later gleaned from various sources is a remarkable set of 300 tanka on hawks, divided into 50 pieces for each of the six categories of spring, summer, autumn, winter, love, and miscellany.

26. Reizei, p. 555.

27. Hiroaki Sato and Burton Watson, From the Country of Eight Islands (Doubleday & University of Washington Press, 1981; reissued by Columbia University Press, 1986), p. 193.

28. Kyoko Selden deciphered for me Jien's notes which are written in hentai kambun.

29. A verse designed to contain all the essential sounds in Japanese, each sound just once.

30. I have translated this sequence acrostically in Anthology 79 (Kobe: Ikuta Press, 1979). In light of Kubota's annotation I would have to revise the translation extensively, correcting many interpretive errors.

31. The sequence of 20 tanka—5 on each of the four seasons—apparently entails some kind of word play; but if it does, Kubota fails to explain—one of the few such spots I have noticed.

32. See the chronological table in Taniyama's Shunzei, pp. 299-302 and p. 336.

33. This rule applies to the ame tsuchi syllabary as well. But that syllabary died out soon. In contrast, the i-ro-ha syllabary has lived to this day.

34. Yasuda Ayao, Fujiwara no Teika kenkyû (revised and expanded) (Shibundô, 1975), p. 91.

35. Teika did the set on the 18th of the ninth month, which is by the lunar calendar the last month of autumn. He did 4 sets of 5-piece acrostics and one 31-piece acrostic sequence (using a tanka, that is), which he remembered as having been composed at Yoshitsune's in the fall of the year.

36. Yasuda's Teika, pp. 428-29; Fukui Kyûzô, Renga no shi-teki kenkyû (Yûseidô, 1969), pp. 13-16; and others.

37. The renga master Bontô (1349-1417) stated that Gotoba told his poets to set the number of links of the renga form at one hundred. See Fukui's book, p. 15. But whatever proof may have existed for his assertion is now lost.