Making Time

Astronomical Time Measurement in Tokugawa Japan

The University of Chicago Press

Contents

Variable Hours in a Changing Society

Hours

What is an hour? The answer to this question in Tokugawa Japan differed considerably from our answer today. First of all, there were only twelve hours in a day. They were referred to as either toki, ji, or koku, and written using a variety of characters. In the English-language literature these hours are often dubbed "double hours," referring to the fact that there were half the modern number of hours in one day. The term "double hours," however, is fundamentally inaccurate, since Tokugawa-period hours were almost never 120 minutes long. This was due to the fact that hours changed their lengths throughout the year. The day was always divided into six hours of light and six hours of darkness. Because the relative lengths of days and nights changed with the seasons, the length of daytime and nighttime hours changed accordingly. If we translate Tokugawa scholars' calculations into our modern temporal units, daylight hours could be as short as 77 modern-day minutes or as long as 156 minutes. Since most of this book deals with Tokugawa-period hours, we will refer to those by the term "hour," and distinguish our contemporary units by referring to them as "Western" or "modern-day" hours.

Of course, most of the people in Tokugawa Japan did not have any particular name for the temporal system that relied on variable hours — it was simply the way they measured time. From later in the eighteenth century, when Japanese scholars began comparing their timekeeping system to the one used by foreigners, they referred to the way they themselves kept time as futeijiho. The literal translation of this term is "method of undetermined hours," which stresses that its hour-units were not bound by a predetermined length. However, as "undetermined" in English might imply a sense of hesitation and lack of resolution, we will refer to the temporal units of the futeijiho day as "variable" hours.

Each of the twelve hours was named according to one of the twelve "animal signs" and given a number that could be announced by striking a bell. Seeing that animal signs may be used for prognostication purposes, they are often misidentified as "zodiac signs" in English, akin to the astrological signs of the Western world that derive from the constellations on the zodiac belt. Animal signs, however, were not related to Japanese zodiac constellations, but derived from an ancient Chinese cosmological system of twelve "terrestrial branches." The "branches" were not originally identified with animals, and therefore, the signs are written with characters different from those used to signify the actual animals. However, by the third century BC in China, the signs came to mean Rat, Ox, Tiger, Rabbit, Dragon, Snake, Horse, Sheep (or Goat), Monkey, Rooster, Dog, and Boar (appendix 1). Even today, traces of this system persist in the Japanese language — the terms for "a.m." and "p.m." in Japanese literally mean "before Horse" and "after Horse."

Unlike the numbers on a Western clock, Japanese hours were not counted by consecutive numbers 1, 2, 3 ... 12. Instead, the midnight hour (Rat) was hour number 9, and the following hours were numbers 8, 7, 6, 5, and finally 4, which was the final hour preceding noon. The noon hour was again numbered 9, followed by 8, 7, 6, 5, and 4, the hour before midnight. Thus, number 9 signified both noon and midnight — the ultimate points of daytime and nighttime. Number 6, on the other hand, referred to the points of change when day turns into night and vice versa. Accordingly, the morning hour of Rabbit was also called the Sixth of Dawn, and the evening hour of Rooster was also known as the Sixth of Dusk (see fig. 1.1).

Wondering "Why do we count the hours the way we do?" several early eighteenth-century scholars investigated classical Chinese texts, and concluded that the origins of this double sequence could be found in the ancient Classic of Changes (the Yi Jing). They came to believe that the system was supposed to represent correspondences between the twelve hours, the twelve months, and the annual cycle of birth and decay. In the Classic of Changes, each stage in the annual cycle is represented using a hexagram. The month of the year with the least daylight is when the dark yin energy is most dominant and the bright yang energy is just beginning its return. This month is represented by the hexagram called "return", in which five broken lines symbolize the plentiful yin and the single unbroken one refers to the weak yang. During the next month, represented by the hexagram "approach", the yang grows stronger, and continues to grow with each month until it reaches its peak in the month before the summer solstice, represented by the hexagram "(full) force". Since the cycle of hours mirrored the cycle of months, hours, too, were associated with the same series of hexagrams. Additionally, each hexagram — and hence each hour — could be discussed in numerical terms. Each unbroken yang line at the bottom of a hexagram was assigned the number 9. Multiplying the number of unbroken yang lines by nine gave the following series: the first hexagram was represented as the number 9 (1 × 9), the second as 18 (2 × 9), and so on: 27, 36, 45, 54. From this series, another series of numbers was derived by taking only the second digit of each number in the first series: 9, 8, 7, 6, 5, 4. Those were the numbers assigned to the hours from midnight to just before noon. Noon, in turn, corresponded to the summer solstice, in which the yin is born and begins its own process of growth. Since the two processes of growth — that of the yang and that of the yin — mirror each other, the hours from noon to midnight were also assigned the same series of numbers from 9 to 4. It is nearly impossible to know whether these early eighteenth-century scholars were indeed right about the origins of the number sequence, nor did they themselves claim their conclusion to be an undeniable truth. They were part of a broader Tokugawa intellectual movement of evidential research (koshogaku) that relied on a philological approach in scrutinizing ancient Chinese texts. As such, they were looking not for metaphysical truths but for historical evidence to explain existing practices.

What Tokugawa-period scholars agreed on was that their temporal system was human-made and used in Japan out of convention. According to major sources from the period, whatever the exact origin of the 9-to-4 sequence was, these numbers were used in Japan by virtue of a conscious human decision. These sources agree that in the seventh century, Japanese emperor Tenchi decided to adopt the Chinese timekeeping system, which at the time used variable hours counted in the double 9-to-4 sequence. Following his decision, Japanese society used this method of hour counting for more than a millennium.

Although there were a variety of technologies available to measure hours of variable length, the most common device employed for this purpose was the incense clock — jikoban (see fig. 1.2). An incense clock was a wooden box filled with sand. Using a template, one would create a channel in the sand and then fill it with powdered incense. Alongside the incense, the user would arrange hour indicators — which looked like little nails — at appropriate distances as signposts to indicate the hours. By watching the progression of the burning incense relative to the indicators one could know the time. Since the indicators were not fixed, it was possible to place them at varying distances according to the season. It was also possible to fill the channel with incense that produced different aromas, so that the change of hours could be detected through the sense of smell.

The use of incense clocks, as well as other nonmechanical timekeepers, did not disappear with the importation and gradual proliferation of mechanical clocks. We find a wonderful description of an incense clock by the Russian naval captain Vasiliy Golovnin, who was captive in Japan from 1811 to 1813 and observed the operation of such a device, at the time still in common use:

[a] relatively small wooden block, covered with clay and whitened; in the clay a narrow channel is drawn, and filled with powder made of some kind of grass, which burns very slowly, and on the sides of this channel there are holes into which one inserts a nail; near those holes there is a designation of the length of day and night hours during the six months from the vernal to autumnal equinox; during the other six months, the day hours become the night ones and vice versa. Thus, the Japanese clock masters find the length of the daytime hour in a certain period, mark it with the nail and, filling the channel with powder, they set it alight at noon, and in this way they measure time. This wooden block they keep in a closed box and try to store it in a dry place.

Seasons

But how did one determine the length of the hours? The system of variable hours might evoke nostalgic notions of "natural" time, attuned to the motion of the sun and the seasonal changes in the length of the day. One might even assume that daily changes in the length of the hours were simply deducted from natural phenomena. Looking closely at the calculations required to maintain such a system, however, we see that it was anything but "natural." First of all, the system of variable hours required a definition of the exact moment that separated night and day. There is no "natural" way to determine this moment. Intuitively, it is very tempting to claim that day begins when the sun appears above the horizon and ends when it disappears below it. But, to mention just one of the many problems with this approach, unless one is at sea, it is difficult to define what the "horizon" exactly is. More importantly, the sky begins to get light long before the sun itself rises above the horizon, and what we refer to as "dawn" is not a specific point but a gradual process that sometimes takes more than an hour. Consequently, in terms of brightness, the day begins at some point before the sunrise. There was an old saying that advised one to look at one's hand: day has arrived if there is enough light to distinguish the three main lines on the palm. This is an obviously subjective measure, dependent on one's surroundings, the weather, and even one's eyesight. For a society to function, there had to be an agreement as to which moment counted as the beginning of daytime.

The beginning of daytime, and hence the length of hours, was defined with reference to an astronomical timekeeping system of equal units. We will discuss how astronomers decided on the parameters for determining the beginning of daytime in chapter 3, but for now let us focus on the temporal units they used before the nineteenth century. In astronomical practice, the day was divided into one hundred equal units called koku, each a little more than fourteen minutes in modern terms. If all hours were of equal length then each hour would have been eight plus one third of a koku, but since hours changed their length with the seasons they were rarely that length. Astronomers determined at what koku the day began and ended. Dividing that interval into six equal parts they then arrived at the length of daytime and nighttime hours for a particular season.

The name of these units — koku — derived from an ancient timekeeping device, a five-vessel clepsydra — a rokoku — that was introduced to Japan from China together with the timekeeping system (see fig. 1.3). Dripping from one vessel to another, there was a steady flow of water that gradually filled the bottom vessel. There, a figurine of a man with an arrow in his hand floated. The tail of the arrow was marked with one-hundred notches. As water dripped into the vessel, the floating figurine rose. By observing how many notches on the arrow's tail were visible above the top edge of the vessel, it was possible to know how much time had passed since the device was calibrated. Reflecting its crucial physical characteristics, the name of the device comprises the two characters "dripping" (ro [TEXT NOT REPRODUCIBLE IN ASCII.]) and "notches" (koku [TEXT NOT REPRODUCIBLE IN ASCII.]). The hundred equal-length "notches," or koku, thus came to signify the basic astronomical division of the day into equal temporal units.

The length of hours was not adjusted on a daily basis. The daily change in the amount of daylight is almost undetectable, and did not amount to even one koku. Nor was the rate of change steady throughout the year, so one could not simply change the length of hours after a predetermined, consistent interval of days in which the daily change in the amount of light amounted to one koku. Instead, people adjusted the length of hours twenty-four times a year according to seasonal units called sekki, each of which measured approximately fifteen days. Each sekki received a name signifying seasonal changes either in atmospheric phenomena or related agricultural activities, such as "great heat," "white frost," "rain on the grain," or "beginning of spring" (see appendix 2). Other sekki names indicated the beginning of one of the "major" four seasons, or important astronomical events such as equinoxes and solstices. Each sekki was further divided into three environment-oriented episodes called ko, which were supposed to represent more specific changes in flora and fauna, such as the appearance of insects after their winter hibernation, the return of migrating birds, or the blooming of various flowers and trees.

Looking at the names of the sekki and ko, it is very tempting to conclude that Tokugawa Japanese were highly attuned to minute seasonal changes and the natural phenomena around them. They may indeed have been attuned to seasonal changes, but this is not something we can deduce from the terms of sekki and ko. The names used in Tokugawa Japan were inherited from the ancient Chinese calendar and therefore, if anything, were supposed to reflect changes in the natural environment of northern China, where the system was created. Consequently, we should not be surprised to find that the list of ko featured animals like tigers, which do not inhabit the Japanese islands, and that the timing of natural phenomena according to ko did not coincide with when they actually occurred in Japan. Not to mention the fact that Japan itself is far from climatically homogenous, which made those names even less indicative of natural cycles in various regions.

People, of course, were not oblivious to these discrepancies. After all, it is not hard to notice that the period called "beginning of Autumn" was the hottest period of the year, while the period "snow turns into rain" could in fact bring sudden snowstorms to areas that do not see much snow during the winter at all. Scholars had already begun to problematize this discrepancy in the seventeenth century, urging the adoption of a different list of ko names that would better correspond to natural events in the Japanese climate. Scholarly debates notwithstanding, the general population was reluctant to part with convention, and kept using the less accurate yet familiar Chinese ko names.

Calendars

There was another incongruity that simply could not be ignored — the incompatibility of solar seasons with lunar cycles. As in many other societies, a month in Tokugawa Japan was defined by the cycle of the moon. But that cycle is 29.53 days long, and therefore cannot be defined in terms of whole numbers of days. To adjust this odd length to practical standards, shorter and longer months were defined, respectively comprising twenty-nine and thirty days. But which months should be short and which long? In our modern-day calendar, short and long months alternate (most of the time). But if one wants to keep months aligned with the lunar cycles, this solution is imprecise. Furthermore, twelve lunar months add up to only 354 days, so if one wants the same months to occur in the same season year after year, a periodic intercalary month is required.

Thus, not only did the system of variable hours — and the luni-solar calendar it relied on — require human regulation, but it required more regulation than the solar calendar we currently use. Since the luni-solar calendar had to adhere to a multitude of requirements, calendrical patterns were based on multiyear cycles. The same sequence of long and short months repeated only every forty-three years. The solar-based sekki repeated their position relative to the twelve lunar months only every nine years. And the position of the intercalary month was based on a period of forty-seven years.