Points of Discovery
The Hill at Nandana
You will have read that al-Biruni climbed a hill in the Punjab around a thousand years ago and measured the Earth to within one per cent. He did climb the hill, and the method he used there was new. But the one-per-cent figure depends on how long you take his cubit to be, his arithmetic and the air both pulled his answer off, and al-Biruni himself said the older figure was the better one.
The myth
Now, you will have met this one in a maths classroom or a pleasant article about the golden age of Islamic science. Sometime in the early eleventh century the great scholar al-Biruni stood on a hill in the Punjab, near a fort called Nandana. He measured the hill's height. Then he measured how far the horizon sank below level. With a little trigonometry he worked out the size of the whole Earth, and he was astonishingly close.
How close depends on who is telling it. A diagram on Wikimedia Commons, drawn in 2010 and copied all over the web, gives his radius as 6,339.9 kilometres, "16.8 km less than the current value". A mathematics blog, That's Maths, notes that he is "often stated" to have come within one per cent. The usual moral is that a lone scholar with an astrolabe matched the modern surveyors.
That story is told in good faith, and the man really did the thing at its centre. So let us take it as a case. We have a hill, two numbers and a famous answer. Our question is whether the answer is as good as everyone says, and whose answer it is.
The people on the hill
It is a small cast, and they lived across eleven centuries.
Abu Rayhan al-Biruni, born on 4 September 973 near Kath in Khwarazm, south of the Aral Sea. Michio Yano, in the Biographical Encyclopedia of Astronomers, says the Sultan Mahmud took him to Ghazna as a political prisoner, and that he stayed in Ghazna for the rest of his life. One small cruelty of history: Yano notes that the old French word aliboron, meaning a fool, comes from his Latin name.
Mahmud of Ghazni, the conqueror who held him. Yano says al-Biruni went with Mahmud on campaigns into India and questioned Indian prisoners, some of them learned men. That is where much of his Sanskrit science came from. It is not a comfortable fact, and it is part of the story.
The caliph al-Ma'mun's surveyors, who worked two centuries earlier in the flat desert of Sinjar, in what is now northern Iraq. According to an account al-Biruni copied out, they brought carpenters and brassworkers along to keep their instruments in repair. They walked north and south until the noon Sun had shifted by one degree. Their answer, as al-Biruni gives it, was 56⅔ miles to the degree.
Mahmoud Bey, a nineteenth-century Egyptian astronomer who tried to fix the length of the old Arabic cubit in metres. Keep him in mind. He matters at the end.
Raymond Mercier, a historian of astronomy, who wrote the chapter on Islamic geodesy for the University of Chicago's History of Cartography in 1992. He is our detective, and he went over the numbers carefully.
What the witness himself says
Al-Biruni tells the story twice: first in the Tahdid, his long book on finding the coordinates of places, written in roughly 1018 to 1025, and again more briefly in his great astronomy, al-Qanun al-Mas'udi, of about 1030.
He did not start at Nandana. He had first hoped to do things the old way, measuring a long north-south line on the plains near the Caspian Sea. Mercier says that plan seems to have failed for lack of support. Then, while he was held at Nandana, he saw that the place itself would do. A hill looked out over a broad, flat plain, and he decided the plain could stand in for the smooth surface of the sea.
The idea is very simple once you see it. From the top of a hill the horizon is not level with your eye. It sits a little below, because the ground curves away. The higher you stand and the smaller the Earth is, the further it sinks. So if you know the height of the hill and the size of that dip, you can work out the Earth's radius. No desert, no camels, no team of brassworkers. One man and one hill. In the Tahdid, Mercier reports, al-Biruni presents it as a method of his own devising, one that spared him the long walk through the desert.
He got the height by sighting the top of the hill from two places on the plain. In Mercier's translation of the Qanun, he gives it as 652 cubits and a little over: 652;3,18 in his base-sixty notation. Then he went up and sighted the horizon. He found the dip a little under 35 arcminutes, a shade more than the width of the full Moon. And then comes a phrase that rewards a second look: "I took it as 0;34." Thirty-four arcminutes, then.
From those two numbers he got a radius of about 12,851,370 cubits. With π as 22/7, that worked out to 56 miles, 5 minutes and 50 seconds per degree, or roughly 56.1 miles.
Here is the part the myth leaves out. Al-Biruni set his own answer beside al-Ma'mun's 56⅔ and kept theirs. His reason, in Mercier's translation, was that "their instrument was more refined", and that they had taken greater pains. He presented his hill as a clever check that roughly agreed with the old survey. He did not offer it as the better measurement.
Turning cubits into kilometres
So how does 56 Arabic miles to a degree become 6,339.9 kilometres? Only if you know exactly how long the cubit was. And that, it turns out, is a whole second case.
Mercier sets out the usual chain. The Arabic mile was 4,000 cubits. In the 1870s Mahmoud Bey, working from old Egyptian and Mesopotamian measures, settled on a cubit of 49.3 centimetres. That makes a mile of 1,972 metres. Mercier adds a detail that matters a great deal. He found that length by other routes first. But part of what persuaded him that this cubit, and not the Egyptian one, stood behind the Arabic mile was that it made al-Ma'mun's 56⅔ miles come out close to the true length of a degree.
Now you can see the trouble. Al-Biruni's hill gave him an answer close to al-Ma'mun's. We then turn his cubits into metres with a cubit that was chosen, at least in part, because it makes al-Ma'mun's answer correct. With that cubit his radius comes to about 6,336 kilometres, about half a per cent short of the modern mean of 6,371. Some of that famous agreement was put there by the conversion.
And the cubit will not stay put. Mercier notes that in the Tahdid al-Biruni calls his cubit the one used for measuring cloth, which may not be the surveyor's cubit at all. Wikipedia, citing a 1973 article by Vibert, says an 18-inch cubit gives about 3,600 miles and a 22-inch cubit about 4,200. Another figure you will find, 3,928.77 English miles, is compared on Wikipedia and in a 2013 paper with a "mean radius" of 3,847.80 miles. That is smaller than any radius the Earth actually has; the real mean is about 3,959 miles. Even the witnesses for the legend cannot agree on the evidence.
The clues that do not fit
Suppose we set the cubit aside and look only at the hill. Three things are wrong, and Mercier found all of them.
The arithmetic. For an angle as small as 34 arcminutes, everything depends on how many digits you carry. Al-Biruni's cosine was very slightly off, far down in the digits. Mercier redid the sum with the correct value and got about 58 miles to the degree instead of 56. That puts the radius over three per cent too large on the usual cubit (that last figure is my own arithmetic). Mercier's verdict is that only by accident did the answer land so close to al-Ma'mun's.
The hill. Mercier thought the most likely peak was one about a kilometre and a half south-southwest of the fort. It stands about 265 metres above the plain. Al-Biruni's 652 cubits would be about 321 metres. If al-Biruni's 652 cubits really describe that 265-metre hill, Mercier notes, the cubit would have to be about 40 centimetres, unlike any cubit known elsewhere. Run his radius through a 40-centimetre cubit and the Earth shrinks to about 5,100 kilometres (my sum again). That is roughly a fifth too small. The physicist John Huth, in a 2013 book I know only through Wikipedia's summary, also puts the result within only about 20 per cent.
The air. Light does not travel in a straight line through the lower atmosphere. It bends a little, following the curve of the Earth, so you see slightly over the edge and the horizon looks higher than it should. Al-Biruni made no allowance for this. Mercier says astronomers of his time did not know about it. From 321 metres, Mercier estimates, refraction would cut the visible dip to about 32 arcminutes. From the real hill of 265 metres it would be about 29. Al-Biruni reported a little under 35. And small errors grow quickly here. The radius depends on the square of the angle, so a single arcminute either way moves the answer by about six per cent (my arithmetic, from the standard formula).
There is also the trouble of seeing anything at all. Mercier reports that the scholar S. S. H. Rizvi climbed the same hill to look for himself. Dust and haze defeated him several times. Only one day after rain could he see the horizon clearly. He did not record a dip.
Put the clues together and Mercier reaches an uncomfortable conclusion, and he states it without hedging. He believes al-Biruni found the observation too difficult, and worked the angle out backwards from his hill height and the known length of a degree. I tried that sum myself: 652 cubits and al-Ma'mun's 56⅔ miles give a dip of about 34 arcminutes, which is exactly what he wrote down. That is suggestive, not proof. It is one historian's reading of the numbers, no document confirms it, and it could still be coincidence. But remember "I took it as 0;34". The case has a loose thread, and an honest narrator has to leave it showing.
Why the truth is the better story
The myth gives us a lone genius on a hilltop who beat modern surveyors to the decimal point. It is a nice picture, and almost none of its numbers can be trusted.
What the record shows is better than that. A man held by a conqueror, far from home, without the support to mount a proper survey, looked at an ordinary hill and a dusty plain and saw a way to size the planet from where he stood. That idea is his, and it is good. The same idea came back in Europe centuries later. Mercier notes that the French astronomer Jean Picard, writing in 1671, warned that refraction makes the trick unreliable. Picard was right, and he was writing six centuries after al-Biruni.
Then look at what al-Biruni did with his own result. He did not claim it was better than the old survey. He compared the two, judged that the caliph's team had the finer instrument and had taken more care, and kept their number. That is not the habit of a showman. It is the habit of a careful observer.
And here is the final turn. The modern legend of his one per cent was built from his hill, a cubit fixed partly by trusting al-Ma'mun's survey, and a slip far down in the digits. In the end it gives al-Biruni credit for a precision he never claimed. The person in the story who was least impressed by the measurement at Nandana was al-Biruni himself.
On sources. The main source for al-Biruni's method, his figures and every reconstruction here is Raymond P. Mercier, "Geodesy", chapter 8 of J. B. Harley and David Woodward (eds.), The History of Cartography, volume 2, book 1 (University of Chicago Press, 1992), read in the press's open PDF. The two quotations from al-Biruni ("I took it as 0;34" and "their instrument was more refined") are Mercier's English, translated from al-Qanun al-Mas'udi, book 5, chapter 7 (Hyderabad edition, 1954–56); I did not read the Arabic. From Mercier also come: the failed plan near the Caspian and al-Biruni's detention at Nandana; the dates of the Tahdid (about 1018–25) and the Qanun (about 1030); the height of 652;3,18 cubits; the dip read as a little under 0;35; the radius of 12,851,369;50,42 cubits and 56;5,50 miles per degree; the corrected cosine giving about 58;11 miles; the conclusion that the close match was an accident; the account of al-Ma'mun's Sinjar surveyors and their artisans, which reaches us through al-Biruni's quotation of Habash al-Hasib; Mahmoud Bey's 49.3-centimetre cubit and its partial basis in the 56⅔-mile degree; the cloth-cubit remark in the Tahdid; the probable peak at about 265 metres and the implied 40-centimetre cubit; the refraction estimates; Rizvi's climb (Mercier citing Rizvi's paper in the 1979 Karachi Al-Biruni Commemorative Volume, which I did not see); the suggestion that 0;34 may have been worked backwards; and the Picard warning. Mercier is the only historian I found who questions the figures in such detail, so his conclusions are one expert's. He says plainly that we know little about al-Biruni's instrument. He states the back-calculation with confidence, but it is an inference from the numbers, not something any document records. My own arithmetic, flagged in the text, covers the radius in kilometres from the 0.493-metre cubit, the over-three-per-cent figure from the corrected cosine, the roughly 5,100 kilometres from a 40-centimetre cubit, the six per cent per arcminute, and the 34 arcminutes worked back from al-Ma'mun's degree. The birth date, the political imprisonment, the Indian campaigns and the questioning of prisoners, the circa-1050 death and the aliboron etymology are from Michio Yano's entry in The Biographical Encyclopedia of Astronomers (Springer, 2007), as hosted by McGill University. The 6,339.9-kilometre and 16.8-kilometre figures are from the description of Nevit Dilmen's 2010 diagram on Wikimedia Commons; it compares with a value near the Earth's polar radius, not its mean. The one-per-cent phrasing and the blogger's own scepticism are from That's Maths, "Al Biruni and the Size of the Earth", 10 June 2021. The 3,928.77 and 3,847.80 mile figures appear in A. C. Sparavigna, "The Science of al-Biruni" (arXiv:1312.7288, 2013) and on Wikipedia. Vibert's cubit range and Huth's 20 per cent (John Edward Huth, The Lost Art of Finding Our Way, 2013, pp. 216–17) were read only through Wikipedia's al-Biruni article. That article also gives the radius as 12,803,337 cubits, which does not match Mercier's recomputation; I have followed Mercier. Not reached: Jamil Ali's 1967 English translation of the Tahdid (an Internet Archive upload exists, but its permission status is unclear, so I did not use it); E. S. Kennedy's Dictionary of Scientific Biography entry and his Tahdid commentary; S. H. Barani's 1951 translation of the Qanun passage; the Encyclopaedia Iranica entry, which refused automated access. The dating of the Nandana observation is known only as falling within the writing of the Tahdid.