Seasonal Living: Five Slots, Four Seasons

Wu xing is a five-part classification, and a solar year is a four-part cycle. Any scheme that puts elements onto seasons therefore has to decide what the fifth slot is, and that decision is not decoration: it is what gives the earth prescription a beginning and an end. Everything downstream inherits those dates — which month to act in, which flavour to favour, which organ to protect.

Two fills are in circulation. One gives earth a season of its own, 长夏 chang xia, "long summer", inserted between summer and autumn. The other gives earth no season at all and instead takes eighteen days out of each of the four, at the end. The first is the version most English-language guides print. The second is the version the medical canon states most explicitly, and it states it as a correction to the first.

《素问·藏气法时论》 assigns each of the five organ systems to a season in one line: 脾主长夏 — the spleen governs long summer. That line is where the fifth season comes from. It is not a later commentary or a Ming-dynasty addition; it sits in the text alongside the four ordinary seasons.

《素问·太阴阳明论》 asks the same question directly and answers it the other way: 脾者土也，治中央，常以四时长四藏，各十八日寄治，不得独主于时也. The spleen is earth, governs the centre, takes charge of the four organs across the four seasons, eighteen days at a time, and does not preside over a season of its own. 不得独主于时 is not ambiguous. It says earth has no season.

The arithmetic behind the eighteen days explains why both lines survived. Four seasons at eighteen days is seventy-two days. A fifth of three hundred and sixty is also seventy-two. The scattered reading therefore hands earth exactly the share it would get as a season of its own — the same total, distributed instead of contiguous. The two lines are not competing over quantity. They disagree about whether earth's days arrive in one block.

The block is where practice diverges. 长夏 as most modern guides use it is the sixth lunar month, about 29.5 days. 长夏 read as a true fifth of the year is about seventy-two. So the window in which earth advice applies runs from roughly twenty-nine days to roughly seventy-three — a factor of two and a half — and both ends can be quoted from the same two chapters. For a reader whose chart is read as needing earth, that is the difference between one month of the year and a fifth of it.

The scheme assumes four equal seasons and carves them up. The calendar does not deliver equal seasons. The boundaries are the four 立, and they are instants rather than dates: in 2026 spring opens on 4 February at 04:02, summer on 5 May at 19:48, autumn on 7 August at 19:42, winter on 7 November at 17:52, and spring opens again on 4 February 2027 at 09:46 — all Beijing time, computed from solar longitude and checkable against any almanac.

Summer runs 5.34 days longer than winter. The cause is orbital: Earth passes perihelion in early January, when it is closest to the sun and moving fastest, so the northern winter is the stretch it covers quickly. This is the same reason the sun's disc looks slightly larger in January than in July. Nothing about elements enters the calculation, but the consequence lands inside the element scheme: eighteen days is 19.9 per cent of spring and 20.3 per cent of winter, so "one fifth of each season" is not a fixed fraction of the year. It is a fixed eighteen days measured against seasons that are not the same length.

That matters for the earth window and for every other fractional rule in the scheme. A fifth, a quarter, eighteen days — each of these is a different instruction once the denominator moves by five days. The difference is small in any single year and accumulates over a life, which is exactly the kind of error a scheme built on ratios should have been designed to survive.

Each of the twenty-four terms is a fixed solar longitude, fifteen degrees apart, with 立春 at 315 degrees. That makes a term an astronomical instant: the same moment everywhere on Earth, computable decades ahead, printable to the minute. It is a coordinate on the ecliptic, and coordinates do not have weather.

Because the coordinate is global it is also indifferent to where you are standing. 立冬 arrives at 17:52 in Harbin and at 17:52 in Singapore. What arrives with it is not the same thing.

The older layer of the calendar knew this and froze it anyway. The seventy-two 候 — three per term, five days each — carry phenological descriptions: 东风解冻, 蛰虫始振, 鱼陟负冰 for the three pentads of 立春. Those are observations of one place in one climate. They were then attached to fixed five-day slots, so the observation and the slot separate as soon as the reader moves north, south, or uphill. A pentad is a coordinate too, and a thaw is not.

The quantity that actually shifts with the season, and that a body has to respond to, is day length and the rate at which it changes. That follows from latitude and solar declination, so it can be computed for any place on earth without leaving the desk.

At 1.3 degrees north the year's light budget changes by ten minutes. At 45.8 degrees it changes by seven hours — forty-four times as much. A seasonal scheme written on the Yellow River plain, between roughly 34 and 40 degrees north, was written for a place where the swing is five to six hours. Applied at the equator it describes a year that does not happen; applied at 50 degrees it understates one that does.

This is the transfer problem in its plainest form. The element label is a coordinate and travels freely. The season it is supposed to name is a local experience and does not travel at all. When a guide says spring is for beginning things, it is quoting a coordinate and asserting a local effect, and the two are only joined at the latitude where the guide was written.

Two charts, both with a 庚 metal day master, sixteen years apart. The first is 1988-06-14 at 07:20 — 戊辰 戊午 庚子 庚辰 — with 土 four, 金 two, 火 one, 水 one and no 木 at all. The second is 1993-01-09 at 05:40 — 壬申 癸丑 庚寅 己卯 — with 木 two, 土 two, 金 two, 水 two and no 火. Both counts come from the four-pillar calculator, stems and hidden stems together, not from adding up characters by eye.

The first was born in a 午 month. By solar term that is summer, the fire season, and the seasonal table puts 庚 at 沐浴 there — metal in a fire month, the weakest reading the table gives it short of 死. Read seasonally, this chart's prescription is a season: autumn if metal needs support, spring if wood does. Either way the advice has dates attached, and the dates are the ones in the table above.

The second was born in a 丑 month. By solar term that is winter, the water season. By branch it is earth. Both labels are correct and they point different ways. Winter says water is strong and fire is dead, which is exactly what the chart shows — 火 absent, 水 twice. The branch says the month itself is earth, which the chart already carries twice. Read the season and this person needs fire, whose season is summer. Read the branch and the month is already earth, so the earth window applies instead, and that window is either the sixth lunar month or eighteen days at the end of every season.

For the second chart the question "when do I act on this?" has at least three defensible answers from the same two chapters: next summer, the sixth lunar month, or the last eighteen days of any season. They are not compatible — they name different months of the same year. The chart did not create that ambiguity. The fifth-slot problem did, and it surfaces whenever the month branch is one of the four earth branches 辰 未 戌 丑, which is four of twelve months.

That is the useful form of the objection. Not that the scheme is arbitrary, but that one reader in three receives an instruction whose start date depends on a choice the scheme never makes explicit. A practitioner who quotes 脾主长夏 and a practitioner who quotes 不得独主于时 will give the same chart different calendars, and neither of them is misreading the text.

The claims in this material split cleanly, and the split is not between tradition and science. It is between claims that name a quantity and claims that do not.

立冬 at 17:52 is a traditional claim and it is checkable to the minute. "Autumn is the season of letting go" is also traditional and it is not checkable at all, because nothing in it can come out wrong. Both arrive in the same paragraph of the same almanac, which is why the whole body of material tends to be accepted or rejected in one movement. It does not have to be.

A seasonal guide that can actually be used states three things: the latitude it was written for, the dates on which its seasons begin and end, and which reading of the fifth slot it adopts. Without the first, the daylight advice cannot be transferred. Without the second, the advice has no edges. Without the third, the earth window silently ranges over a factor of two and a half and the reader cannot tell which one they were given. Those three are cheap to supply, and they turn a mood into a schedule. A schedule can be checked against what happened; a mood cannot, which is why it survives every test.