First Snow on Daffodils: Writing Reality

First published in Nesting Dolls, the 2018 members’ anthology for the Yuki Teikei Haiku Society, pages 63 to 72. Newly added here is the paragraph that begins “It is thus no surprise,” the translation that follows, and then the following sentence of commentary. Originally written in December of 2014, and revised in 2016, 2017, and 2018. My thanks to Philip Kennedy for editorial suggestions and refinements. See also the new postscript at the end.       +

In Bashō and His Interpreters (Stanford, California: Stanford University Press, 1991), Makoto Ueda presents the following Bashō poem, with commentary (149; Ueda’s book does not include the Japanese, but does include the romaji and transliteration). It’s a poem that initially puzzled me:

                the first snow

                just enough to bend

                the daffodil leaves


                hatsuyuki ya suisen no ha no tawamu made

                first-snow | : | daffodil | ’s | leaf | ’s | bend | till

Could this really happen? Winter’s first snow typically falls in late autumn or early winter. Daffodils bloom in early spring, and their stalks and leaves have usually decayed to mulch by early summer. Their leaves are long gone by late autumn or winter. The first snow and daffodils would seem to be mutually exclusive, seasonally, yet here we have them together in this poem by Japan’s greatest haiku master. Did Bashō make this up? Is the poem authentic? How are we to make sense of it, and what can we learn from this haiku as a result?

        Let’s look more closely at the poem. In Japanese, as Ueda indicates with the transliteration that accompanies the romaji, “yuki” is the word for “snow,” and “hatsuyuki” is “first snow,” a well-established season word. The word “suisen” is “daffodil” or “daffodils” (and also “narcissus”; daffodils are in the narcissus family), so the combination of snow and daffodils is inescapably intentional, and not a translation error. But are the daffodils perhaps not in bloom? The word “ha” is “leaf” or “leaves” in Japanese, which might suggest just the daffodil’s stalks, long after the flower has bloomed. If these stalks or leaves might somehow remain all summer and autumn, could winter’s first snow fall on them then, at the start of winter? Or could these daffodil leaves have just emerged from the ground, very early in spring, before the appearance of their renowned bright-yellow trumpets?

        Neither is the case, however. Bashō created a haiga for this poem, and his painting, reproduced in Ueda’s book (148), shows flowers that are blooming. Snow can happen at various times of the year, but daffodils bloom only in spring, ahead of tulips, which would suggest that this is a spring poem, or perhaps late winter. It can snow in spring, of course, but how could that be the season’s first snow, and therefore late autumn or early winter, as seemingly intended by that seasonal reference? How can we resolve the blooming of daffodils in spring with the winter’s first snowfall?

        One subtlety here is that “suisen” can be translated as “narcissus,” which bloom earlier than daffodils. William Higginson’s international saijiki, or season-word almanac, Haiku World (Tokyo: Kodansha International, 1996), translates “suisen” as “narcissus,” and offers further insights. Higginson writes that “In haikai narcissus (suisen) by itself always means late winter” and that “it seems best to accept the Japanese tradition of narcissus as late winter and daffodil as mid spring” (96). This observation might suggest that “suisen” could be better translated as “narcissus” instead of “daffodils,” even though Higginson’s comments in Haiku World come from the discussion of daffodil as a spring season word. Even if “narcissus” might be more accurate for Bashō’s poem, however, that still does not resolve the problem of how a first snow could fall when the narcissus blooms.

        In Bashō: The Complete Haiku (Tokyo: Kodansha International, 2008), Jane Reichhold translates this poem as “first snowfall / enough to bend down / narcissus leaves” (91), and assigns the poem to winter. She refers to Bashō’s haiga, in which “narcissus [not daffodils] are shown with bent leaves” (283). She says “The traditional practice would have been to associate the whiteness of the flowers [these are apparently not yellow daffodils] and the snow,” and adds that “Bashō took his own advice and wrote about the less showy aspects of the scene” (283). Likewise, one commentator in Ueda’s book notes that “Bashō’s own painting that accompanies the hokku shows white flowers, too” (149). However, are the flowers themselves white, or is it the snow that makes them look white? (I think of Richard Wright’s poem about a presumably black boy holding out his hands in falling snow until they turn white.) If the flowers in this poem are narcissus, they would be white—in which case Ueda’s translation would seem to benefit from using “narcissus” rather than “daffodils.” Daffodils can be white, too, but if they are yellow daffodils, we can interpret the poem as having the snow turn the blossoms white.

        In Bashō’s Haiku (Albany, New York: State University of New York Press, 2004), David Landis Barnhill also translates this poem with “narcissus” rather than “daffodil”: “first snow— / just enough to bend / narcissus leaves” (56). In his notes on the poem, Barnhill assigns this poem to winter, listing both “first snow” and “narcissus” as the poem’s winter season words (185). Barnhill also identifies “suisen” as one of the “major nature images” in Bashō’s haiku, providing a gloss on the term: “Chinese narcissus. Blooms white in late winter, suggesting purity, with leaves up to sixteen inches long” (275). Thus the purity of snow would be an attractive association with the purity of white narcissus blooms. An early lesson we can learn here is that translation is a difficult game. I usually find Ueda’s translations to be faultless, but here I would lean more toward “narcissus” rather than “daffodil” as a translation for “suisen,” for the reasons already mentioned, and for reasons to come.

        According to Gabi Greve’s World Kigo Database, as we might expect, “hatsuyuki” or “first snow” is listed an early or mid-winter season word (, at least in our current solar calendar. She also says of “daffodils, narcissus and jonquils” that “we have them in late winter and mid-spring,” and lists “suisen 水仙 narcissus” as late winter ( Yet the translation she offers (not her own, but I believe by Eri Takase) refers to daffodils, not narcissus: “The first snow, / Just enough to bend / The leaves of the daffodils.” Higginson’s Haiku World designates “first snow” as a mid winter season word (246) and “daffodil” as a mid spring season word (96). Thus, we can see that the seasonal reference in this poem is complicated. If a first snow has fallen on a blooming daffodil—or narcissus—we may still wonder how. Has Bashō invented this poem, writing from unchecked imagination, creating a poem that lacks authenticity? Readers may still be puzzled: Indeed, how could the first snow fall on daffodils or narcissus? Further complicating matters is the translation in Zen Haiku by Jonathan Clements (in two lines) as “The first snow / Just enough to bend the jonquil leaves.” Higginson does not list jonquils in his Haiku World saijiki, but of course they are part of the narcissus family. Even though narcissus may be a better translation than jonquils, too, we are still left with the puzzle of how the first snow could fall on any of these flowers.

        Yet another clue is provided in Ueda’s book, with a notation that “The first snow of the winter fell on January 31” (149). Because the date is so specific, it could be that this note was written with the poem as an explanation that this was when that year’s first snow actually fell. This turns the poem’s reference to a “first snow” into a mere fact (which could be at various times of year) rather than being a season word (which we would expect to focus on when it typically occurs, in late autumn or early winter). Other notes in Ueda’s book all seem to be Ueda’s comments, though, rather than being the poet’s comments or headnotes, so it could be that Ueda is providing context that he knows from another source, rather than this being Bashō’s note. In contrast, the notes to this poem in Barnhill’s book say “1686–87 (18th of Twelfth Month; January 31, 1687)” (185), which sounds like a known date associated directly with the poem rather than being any kind of extraneous note by the translator. In any case, we still may be at sea in trying to resolve this poem’s factuality.

        Let’s dig a little deeper. Barnhill’s note provides a date of seeming composition (or the date that the poem was about), but does not mention that this date happened to be when the first snow fell that year. Ueda’s note says that January 31 was when the first snow fell, and is the only direct exploration of this poem’s timing in his book. The commentary Ueda includes doesn’t even raise the question of how a “first” snow could fall on daffodils. This suggests that commentators saw no issue with the seasonal content as Bashō presented it. They did not have my doubts. It seems to strike them as completely factual and believable. Indeed, they talk about how delicate the snowfall must be—just enough to bend the leaves, but not flatten them—and that the poet must have seen this image with “the innocent eyes of a child” (149). One commentator even says, “There is something fitting about the combination of the green leaves and the snow. Daffodil leaves and first snow go well together” (149). We see no hint of puzzlement in any of these comments, or of trying to resolve any seasonal conundrum. To Bashō’s commentators, this poem has no puzzle to resolve. But still, how could a first snow fall on daffodils?

        One answer may be obvious: The first snow did not fall in the autumn or even early winter at all, and didn’t come until the daffodils were already starting to bloom in spring (or perhaps late winter). Thus, one way to resolve this poem is to understand that “first snow” is not functioning as a season word, but is simply a fact of reality, and happened to occur at a much later time than we would normally expect. The reference to “first snow” therefore seems unfaithful to hon-i, or poetic essence, which would employ season words when we usually expect them to happen, a trait that impacts the seasonal allocation of surely all Japanese season words. Rather, Bashō can be seen to have written about reality as he experienced it, feeling no need to be bound by whatever classifications a saijiki might offer regarding first snowfalls and when they’re supposed to happen. Bashō did not hesitate to use the phrase “first snow,” even though, according to every saijiki, it would seem to be at odds (as a season word) with daffodils. The point to learn here is that while many terms and phrases may be listed in a saijiki as season words, these terms and phrases in haiku do not always function as season words. In this case, the springness of “daffodil” would seem to trump the typical seasonal association of “first snow.”

        But there is more to the poem than this, something less apparent, at least to Westerners, and it lies in the fact that “first snow” actually is the season word, without trumping “daffodils” (or “narcissus”), or needing to trump it as an additional season word. Indeed, the seemingly obvious resolution just discussed is not entirely accurate. Rather, another way to interpret this poem, and the “first snow” phrase in particular, is in terms of the Japanese new year. In the old lunar calendar, which Bashō used, New Year occurred at the end of winter, followed immediately by spring. The lunar new year (also the new year in Chinese, Korean, and Vietnamese traditions) fell on the fifteenth day of the first month. By our current Gregorian (solar) calendar, this date can vary from late January into late February. With this in mind, it is easy to see “first snow” not as the first snowfall of the winter season, but the first snowfall of the new year.

        It is thus no surprise, then, that Sam Hamill translates the poem as follows, from The Poetry of Zen (translated and edited by Sam Hamill and J. P. Seaton, Boston: Shambhala, 2007, 147):

                New Year’s first snow—ah—

                just barely enough to tilt

                the daffodil

Hamill’s version goes straight to telling readers that this is the New Year’s first snow, and also includes “ah” where the “ya” cutting word appears in the original Japanese, something he does not do with many of his other translations, but does so here seemingly to emphasize a sort of delight in the New Year’s first snow.

        For context, it is useful to know that haiku poetics have traditionally designated New Year as a fifth season, and that a typical saijiki in Japanese categorizes poems into five seasons (Higginson’s Haiku World also does this). In the lunar calendar, this fifth season falls between winter and spring, not in the middle of winter as it does in our current Gregorian (solar) calendar. In celebration of Japan’s most prominent holiday, many poems indicate the new year by referring to “firsts”: the first dream, the first sunrise, first letter-writing, first calligraphy, or first laughter. So one way to interpret “first snow” is indeed not as the first snow of winter, but as the first snow of the new year. In other words, it might have snowed just the day before, but today’s snow is still the first snow of the new year. In this context, Ueda’s reference to the “first snow” falling on January 31 makes complete sense. On the topic of “New Year,” Higginson writes that “Despite the dislocation in time [because of moving from the lunar calendar to the solar calendar, after Bashō’s time], some phrases including the word ‘spring’ are still considered New Year topics” (288). This helps to resolve the factuality of a “first snow” falling on daffodils. Indeed, Bashō’s “first snow” poem is correctly apprehended as simultaneously being a first-of-the-year and late-winter poem. White daffodil or narcissus flowers have risen and bloomed, and Bashō goes beyond the surface observation of comparing the whiteness of the flowers to the whiteness of the snow by pointing out the delicacy of exactly how much snow had fallen—just enough to bend the leaves.

        One point to learn here is that it helps to be aware of the lunar calendar in apprehending certain Japanese haiku. Higginson’s Haiku World, and its companion book, Haiku Seasons (Tokyo: Kodansha International, 1996), are particularly helpful in addressing this topic. Another point is that cultural contexts for Japanese haiku lie not only in allusions to famous places or other famous poems, but even in a different way of perceiving nature and the seasons. Japanese seasonal awareness, in fact, is heavily influenced by Chinese seasons, which have been split into as many as seventy-two distinct segments (Liza Dalby writes wonderfully about this in her book East Wind Melts the Ice: A Memoir through the Seasons, Berkeley: University of California Press, 2007, also succinctly summarized at; see more about the seventy-two seasons at and note how these two lists do not seem to agree).

        Yet another lesson to learn relates to the notion of having two season words, called kigasanari (“kigo layering”) in Japanese. Haiku poets are cautioned against using two season words in haiku. This is generally good advice, especially for beginners, if one of the two references is truly redundant or dilutes the poem’s focus. This stance is reflected in the Yuki Teikei Haiku Society’s annual Tokutomi haiku contest, which enforces a strict rule calling for just one season word in each haiku. However, this advice does not mean that one should never use two season words when you have good reason. Indeed, the masters did it frequently. This is evidenced not only in Bashō’s “first snow” poem, but in many other poems as well. In Stephen Addiss’s Haiku: Anthology of Poems (Boston: Shambhala, 2009), I count at least twenty poems by Japanese masters with two season words—sometimes more than two. Higginson addresses the issue in a short section of Haiku World by saying “Whichever season word dominates the seasonal understanding of a poem, and thus its placement in the saijiki, is said to be the season word of that poem” (33) and that “When season words relate to topics in different seasons, usually one or the other obviously governs” (34)—and he emphasizes that this applies equally to classical Japanese haiku and contemporary Western haiku today. Thus, the seasonal focus in a poem about a frog in moonlight is governed by the frog (spring), not the moon (typically autumn). And we would be impoverished if we were to forbid ourselves to write about frogs in moonlight out of a fear of having two season words.

        Thus, again, it is not a property of certain words to be seasonal, but a function. This would suggest that no term is automatically a season word in a given poem just because the same word is a season word in another poem, or listed as a season word in a saijiki (which is typically intended to be descriptive rather than prescriptive). Indeed, a property is immutable, whereas a function is not, which allows some words to not actually function as season words in some haiku, while they might in others. For example, in the phrase “snow globe,” the snow is artificial and thus not actually a winter reference, even if we might not be able to resist thinking of actual snow. Indeed, words will still tend to take on the seasonality associated with them, but careful readers will look beyond such automatic associations if other aspects of the poem require a deeper understanding. An example of this is my poem, “an old woolen sweater / taken yarn by yarn / from the snowbank.” A superficial reading would take this to be a winter poem, but it is really a spring poem, because birds are snatching bits of yarn from the old sweater to build a nest. Higginson recognized this, and used the poem in Haiku World in the “bird’s nest” category, which he assigns to “all spring” (75). And note that he includes the poem in that category even though it doesn’t actually use the season word itself, which suggests that a season word could be entirely implied, and need not always be used directly. We see this potentiality of looking beyond automatic associations in Bashō’s “first snow” poem, where “daffodil” clearly governs at first, functioning as the season word, whereas “first snow” seemingly does not, even though a deeper reading of the poem reveals that these seasonal references are in fact compatible rather than seasonally impossible.

        Still on the subject of two season words, cases occur where two terms refer to the same season, and are repeated just for the sake of developing an effective expression. An example from Addiss’s book is this poem by Chora (23), which also shows season words from seemingly conflicting seasons:

                         Highlighting the blossoms,

                clouded by blossoms—

                         the moon

Here we actually have three season words. Normally the moon is considered autumn, as already mentioned, and needs to be modified (such as by saying “spring moon”) to make it clear that one means the moon in a season other than autumn. But such a clarification is not necessary in this poem because the blossoms (understood in Japanese haiku to be cherry blossoms) are clearly spring. So this is a spring moon. It would not be logical to have blossoms in the autumn, so this is not an autumn poem. The blossoms trump the moon, seasonally, making this is a spring poem, and making it unnecessary to modify “moon” to indicate that it is not an autumn moon. And of course, the word “blossoms” is deliberately repeated, not just for poetic effect, but because the repetition is essential to the poem’s meaning—the blossoms are lit by the moon and the blossoms hide the moon from the viewer below. Thus the beauty of both the moon and blossoms is emphasized. The poet does not shy away from using two season words—or in this case, three.

        Here is another lesson to learn, that we need to be true to reality, and not be limited by codified restrictions in any friendly neighbourhood saijiki. In his entry for “daffodil,” with no reference to the seemingly problematic Bashō poem, Higginson offers an admonishment: “When writing haiku or hokku it is better to write naturally of what one observes than to worry too much about seasonal associations” (96). He also writes in his introduction, more forcefully, that “Blinding oneself to the actual phenomena of a given place and time because of some loyalty to the saijiki will only interfere with both creating poems and appreciation of the phenomena themselves” (28).

        Ultimately, Bashō’s poem is not problematic at all—even if its clarity isn’t immediately evident to Westerners. If we understand the lunar calendar, and take the first snow to be the first of the lunar new year rather than the first of winter in the solar calendar (potentially three or four months earlier), and consider “narcissus” as a better translation than “daffodil” for “suisen,” then this poem is not built on two conflicting season words at all. Rather, they complement each other. The “daffodils” (or “narcissus”) season word does not trump “first snow,” nor does “first snow” need to trump the flower. Rather, the poem’s two season words are both happening at the same time, around the time of the lunar new year, and the poet mentions both because they are both essential to what he is writing about. Bashō’s poem is indeed authentic, and he has not shied away from writing about what really happened.

        Authenticity, mind you, may be best apprehended as an effect of the poem upon readers, and not defined by whether something “really happened” or not. The “reality” of what happened is never provable, anyway, and a poem may still be effective and authentic even if imagined or put together as a pastiche of experiences or images. The point here is that Bashō did write about what really happened, but “what really happened” may seem unbelievable to us today, because of differences between the solar and lunar calendar. The larger point is that the two season words in this poem are essential, but in actuality, only one of the two potentially seasonal terms is functioning as the season word. Bashō has written his poem regardless of any knee-jerk application of a rule to have just one season word in haiku. Westerners might wrongly presume that this poem’s seeming use of two season words are conflicting, but even if this poem is about snow and daffodils happening when a saijiki might say they should happen, Bashō is nevertheless not afraid to write about things that really happen, even if he has to use more than one season word. We have much to learn from even the simplest of Bashō’s many poems, but a central lesson is that he was writing reality.

Note: The Japanese text for Bashō’s “first snow” poem is not included Bashō and His Interpreters. It is quoted from Yamamoto Kenkichi’s Bashō zen hokku (Kōdansha gakujutsu bunkō, 2012), page 230, and in Kira Sueo and Satō Katsuaki’s Bashō zen kushū: Gendai goyaku tsuki (Kadokawa gakugei shuppan, 2010), poem #887.


Although this essay asserts that Bashō was “writing reality” with his poems, was that always the case? In Traces of Dreams: Landscape, Cultural Memory, and the Poetry of Bashō (Stanford, California: Stanford University Press, 1998), Haruo Shirane points out that Bashō seemed to have no hesitation in rearranging or compressing the sequence of events in his Oku no Hosomichi travel diary, or even saying that a different person wrote a particular verse in a renga. His own “old pond” haiku originally had a different first line and its eventual first line could be said to have been “manufactured” rather than being part of the original experience. Indeed, Bashō routinely revised for literary effect, setting a higher value on aesthetics than on “what really happened.” This behavior flies in the face of the too-common Western sense of haiku as a Zen art that accepts experience and spontaneously depicts it strictly as it is (“first thought, best thought,” and so on). But does that mean Bashō was not writing reality? I would suggest that he still was, because his revisions still came across to readers as authentic and believable—such poems (though perhaps few) still seemed to be reality. He did not alter reality to write about impossible things, but apparently altered reality only to the extent that the scene depicted in the poem was still possible. In the case of the daffodil poem discussed here, what at first might seem to some readers as an impossibility, and thus seemingly invented, turns out to be completely accurate to experience after all. And yet Bashō did not limit himself to “actual” experience. In Bashō’s poetry and prose we find an invigorating range to how he was “writing reality.”

—14 December 2018