Termites · heat · a number that travelled

The Mound That Breathes by Day

The story says termites hold their mound at exactly 87 °F while the veld outside swings from 35 to 104, and that a Harare office block copied the trick and uses a tenth of the energy. Put the published measurements under it and a different mound appears. The nest is steady through the day for the reason a cellar is, and drifts with the soil through the year. The air moves because the day's heat reaches the thin outer walls before the thick core, which flips the draught twice a day, and an abandoned mound with no colony inside does it too.

The number that travelled

On 13 February 1997 the New York Times ran a piece on a new office and shopping block in Harare, the Eastgate Centre, designed by Mick Pearce with engineers from Ove Arup. It explained the building by explaining the termite:

“Deep inside, the insects farm a fungus, their only food. It must be kept at exactly 87 degrees, while the temperatures on the African veld outside range from 35 degrees at night to 104 degrees during the day. They do it by venting breezes in at the base of the mound, down into chambers cooled by wet mud carried up from water tables far below, and up through a flue to the peak.”Donald G. McNeil Jr., “In Africa, Making Offices Out of an Anthill”, New York Times, 13 Feb 1997 (read from an archived copy)

No measurement or scientist is cited for either number. A few paragraphs later the same article describes Harare itself: days as warm as 88 degrees commonly drop to 58 degrees at night. That is a swing of 30 °F. The veld line is a swing of 69 °F, from 1.7 to 40.0 °C, and the German weather service's table for Harare-Kutsaga gives the station's absolute maximum over 1941 to 1976 as 35.0 °C.

The paragraph then travelled through the next quarter century, often nearly word for word:

1. 2001 Tzonis, Lefaivre and Stagno, Tropical Architecture (reprinted by the AIA, 2003): termites must live in a constant temperature of exactly 87 degrees (F) to survive.
2. 2004 A design-firm weblog, linking a forwarded copy of the 1997 article as its source, rewords the vents as constantly opening and closing a series of heating a cooling vents (sic).
3. 2007 Inhabitat, the same sentences, exactly 87 degrees F. Still live.
4. 2008 The Biomimicry Institute, as quoted that year by the biologist J. Scott Turner and the engineer Rupert Soar: termites maintain the temperature inside their nest to within one degree of 31 °C, day and night.
5. 2009 Pacific Standard: daily temperature swings of up to 70 degrees. GreenBiz: a daily fluctuation from 40 degrees C to less than 0 degrees C.
6. 2026 Wikipedia's Eastgate article still quotes the 35 degrees at night and 104 degrees during the day, in the same article as its own typical daily temperature swings between 5 and 14 °C.

The energy number travelled the same road and grew on the way. The 1997 article said the building used less than 10 percent of the energy of a conventional building its size. The engineers' own report, in the Arup Journal the same year, had measured it: 9.1 kWh/m² against 11 to 18.9 in six other Harare buildings, an energy consumption per unit area of 48%-83% of theirs: 48% to 83%, from early readings. The ten per cent that does appear in that report is a different quantity: the cost of the Eastgate installation was about 10% of a full mechanical air-conditioning system. Pearce's own site later said 35% less energy than the average of six conventional buildings, and elsewhere on the same page less than 50%. Arup's current project page says 90% less energy to heat and cool.

None of this makes Eastgate a failure. It is a good building that runs without air conditioning, by storing the day's heat in heavy concrete and flushing it with fans at night. Turner, who studied the mounds for decades, put it this way to a journalist in 2013: Eastgate succeeded because Pearce was a very good architect rather than a crude imitator of nature. What is wrong is the termite in the story, and the tenth.

What a metre of dirt does to a day

Heat soaks into ground as a wave that shrinks and falls behind as it goes. Fourier worked out the rule in the 1810s and published it in the 1820s: a cycle of period P loses a factor of e (to 37%) and falls one radian behind for every penetration depth d = √(κP/π) it travels, where κ is how easily the ground passes heat. A year is 365 times longer than a day, so its wave reaches √365, about 19 times, deeper. The same soil that wipes out the afternoon passes most of the summer.

At half a metre, where Ocko's nest logger sat, ordinary ground keeps between 0.4% and 2.7% of the day's swing for any κ from 0.3 to 0.7. The measured nest kept 1.7% of the flanks' swing: its daily wave is 0.10 °C, a fifth of the logger's stated 0.5 °C accuracy, and the whole plotted trace moves only 1.2 °C, most of it a slow drift across the day. Dirt alone is enough to explain why the nest is so steady through the day; nothing here requires the termites to regulate it. (The nest does run warmer on average than the rest of the mound, and in another species Korb and Linsenmair credit the colony's own heat for that; warmth and steadiness are different questions.)

Fourteen to thirty-one

Through the year the story fails the other way. Turner and his colleague Marais left temperature loggers in two nests at Omatjenne, in northern Namibia, from February 2004 to January 2005. In an unpublished manuscript they report the nest of mound OM123 running from 14.5 °C in June to 31.0 °C in November, a span of 16.5 °C, and the nest of OM126 from 21.0 to 31.5. Their averages over the quarter around each solstice spanned 9.4 °C and 6.4 °C. Turner and Soar's 2008 paper plots what appears to be the first of these nests (from about 14oC in winter to more than 31oC in the summer) and concludes:

“There is just one problem: there is no evidence that termites regulate nest temperature. Indeed, there is good evidence that they do not.”J. S. Turner and R. C. Soar, “Beyond biomimicry: what termites can tell us about realizing the living building”, I3CON, Loughborough, 2008

Run the 2004 surface through the thermometer above and plain ground predicts a solstice-quarter span of 3.6 to 9.3 °C for a nest between half a metre and two metres down, at κ from 0.3 to 0.7. OM126's 6.4 sits inside that band. OM123's 9.4 sits just above its top, as if that nest were shallow, or sat under a patch of ground that heats more than the open-field value the satellite record gives. The nests' depths are not given, so this is a range, not a fit; but both are close to what plain soil does somewhere in that range, and Turner and Marais themselves found nest temperature tracks soil temperatures closely.

What moves the air

The older explanation, Martin Lüscher's in Scientific American in 1961, had the colony's own heat as the engine: warm, wet air rising from the nest up a central chimney, circulating out to the surface to breathe, and sinking back. Turner and Soar put the colony's output at roughly 100 watts. Turner's own tracer-gas work in 2001 found no such circulation and pointed at gusting wind. Then, in 2015 and 2017, Hunter King, Samuel Ocko and L. Mahadevan put a flow sensor inside the conduits themselves, in India and then in Namibia, with Turner and Soar among the Namibian co-authors.

What they found is in the day above. In these mounds the surface conduits sit behind a porous covering 1 to 3 cm thick (Turner, 2001), so the flanks follow the day closely, peaking in late afternoon. The centre sits behind far more material and peaks about 2.5 hours later. Averaged over the four sides, the flanks overtake the centre at 09:23 and fall back below it at 17:27. By day the flanks are up to 1.7 °C warmer, so their air is lighter and rises; at night they are up to 5.1 °C colder and their air sinks. The measured flows agree: in Namibia, air flows strongly down the conduits at night, and, on average, upward during the day, and in India, in mounds of another species standing mostly in forest shade, an abandoned mound showed the same gradients and flows as a live one (though its night flows were weaker), which the authors say rules out metabolic heating as the central mechanism. The mound is ventilated by the day's warming and cooling, twice a day; the colony does not have to drive it.

A difference of 1.7 °C over a metre of height pushes with about 0.06 pascal. The authors' own closed-pipe estimate from differences of 3 and 6 °C comes out near 35 and 70 cm/s, and both papers say that is about ten times what they measured, likely because real conduits are narrow and tangled. The measured flows were of order centimetres per second.

The centre is too lively for dirt

One more thing falls out of the same curves, and it is this page's own reading, not the paper's. In flat ground a buried wave pays for its delay in size by an exact rule: fall one radian behind and you keep e⁻¹ of the swing. The mound's centre falls 2.5 hours behind the flanks, 0.656 radians, so conduction through flat ground would leave it 52% of their swing. It keeps 67%.

A body warmed from every side does better than flat ground, because heat arrives from all around at once. For the same delay the axis of a long cylinder keeps 90%. So the measured 67% could be conduction, but only if the heat crosses very little solid: for the delays to match, somewhere between 6 cm (flat) and 16 cm (cylinder) of material at κ from 0.3 to 0.7. The authors describe the inside of these mounds as a network of broad conduits (∼2–10 cm diameter …) that connects the center with the periphery very well. On that reading the centre's warmth arrives mostly through air, which is the same convection the flow sensors measured. Across the plausible κ of 0.3 to 0.7 the required thickness stays within about a hand's width.

What stays true

Mounds are not nothing, thermally. The nest is far steadier than the air over a day, and in West African Macrotermes bellicosus Judith Korb and K. Eduard Linsenmair reported that Only large colonies attain constant nest temperatures of 30 °C, with the colony's own metabolism supplying part of the warmth, and that Mound structure alone resulted in a relatively constant nest temperature. In an earlier paper (1998) they found that mounds in shrub savanna and in gallery forest are built differently. How much termites manage temperature is still argued over, species by species. What the measurements do not support is the story as told: a nest held at exactly 87 degrees against a 69-degree daily swing, by insects opening and closing vents, copied by a building that uses a tenth of the energy.

-- what the readings say (verify.mjs, section 4)
sides:   mean 26.42 C, daily swing ±5.59, warmest 16:46
centre:  mean 28.42 C, daily swing ±3.75, warmest 19:16
nest:    mean 29.65 C, daily swing ±0.10 (logger accuracy 0.5)
sides minus centre crosses zero 09:23 (up) and 17:27 (down); range -5.14 to +1.68
centre keeps 67.0% of the sides' swing, 2.51 h later; flat conduction: 51.9%
-- the year (section 5)
2004 surface, solstice quarters 28.65 / 17.40 C; plain ground at 0.5-2 m: span 3.6 to 9.3
measured nests 9.4 (OM123) and 6.4 (OM126)