Stonehenge and the megalithic yard

A monument that changed its mind

Begin outside the famous stones. Around 3000 BCE, people dug the circular ditch and built banks, with 56 pits, the Aubrey Holes, just inside. Cremated human remains connect this early monument with burial. What originally stood in the holes remains debated. The great central stone settings arrived around 2500 BCE; the bluestones were rearranged later, and the Avenue connected the monument with the River Avon. The Y and Z pits belong to still later activity, dated by English Heritage to about 1800–1500 BCE.¹

The name henge concerns an earthwork enclosure; the stone circle is a different architectural element. Putting every pit and stone into one diagram is useful for orientation, but it can conceal centuries of change. A convincing unit theory must say which features belong together and why a measure would persist between their construction episodes. A shared modern map does not establish a single prehistoric design.

The craftsmanship is much less elusive. The sarsen uprights and lintels have mortice-and-tenon joints; adjacent lintels connect with tongue-and-groove joints. English Heritage’s account of the stoneworking explains the careful finishing, including the inward-facing surfaces of the trilithons.⁴ Those connections demanded coordination of shape and position. They do not require every dimension to be an integer in the same yard.

What Thom was actually proposing

Alexander Thom’s proposal was more ambitious than finding an attractive ratio at one site. He argued for a recurring length behind megalithic plans. The familiar yard is about 2.72 modern feet, or 0.829 metres; his megalithic rod is 2½ such yards. The words yard and rod are modern analytical names here, not recovered Neolithic terminology. Porteous’s contemporary critique identifies the central claim as a remarkably consistent unit across sites.⁵

In their 1974 Stonehenge paper, Alexander Thom, Archibald Stevenson Thom and Alexander Strang Thom presented a new survey made in 1973. Their plans distinguish stones considered undisturbed, straightened or re-erected. They preferred the worked inner faces when reconstructing the sarsen geometry. One proposed design placed the ring between circles with circumferences of 45 and 48 megalithic rods.²

There are two kinds of information on such a plan. The observed positions constrain what survives. Superimposed circles, ellipses and integer counts propose what the builders intended. The latter can be tested, but they cannot be treated as another set of independently measured dimensions. A reconstruction that starts with a yard will naturally contain dimensions expressed in that yard.

To see the decisive step without the complexity of the central stones, move to the Aubrey ring. It supplies an unusually clear worked example.

How one circle becomes a ruler

Thom’s team report a mean Aubrey-ring radius of 141.80 ± 0.08 modern feet, with a standard deviation of individual radii of 0.56 feet. They derive a circumference of approximately 891 feet and then assume an intended circumference of exactly 131 megalithic rods.² The radius belongs to their survey analysis. The 131-rod count belongs to their design interpretation.

That is a close match to the proposed yard. But the inference depends on the assumed count. The same reported circle can produce these exact numerical fits if we choose different counts:

Only the first count is attributed here to Thom’s Aubrey-ring interpretation. The other three are our counter-examples, chosen to expose the arithmetic, not rival archaeological discoveries. None requires moving a hole: the physical circle stays fixed while the assumed number of units changes.

This does not refute the yard. It demonstrates why one circle cannot uniquely identify it. A radius, its doubled diameter and its calculated circumference are three expressions of one size, not three independent confirmations. A shared-unit claim becomes stronger when a unit fixed using one set of evidence predicts measurements in a genuinely separate set.

The quoted errors also describe different things. The 0.56-foot radial scatter is about 171 mm; the reported ±0.08 feet on the mean is about 24 mm. A well-estimated average does not mean that every hole was placed within that smaller interval. If we hold the 131-rod count fixed, propagating the reported ±0.08 feet gives roughly ±0.0005 m on the inferred yard. That calculation describes uncertainty within the chosen model; it does not establish the count or eliminate uncertainty about ancient intention.

The alternative was never simply “random stones”

A statistical quantum is a repeated increment inferred from a collection of lengths. One searches for a candidate that brings many observations close to whole multiples. That is a sensible question. The harder question is what comparison would make a successful fit persuasive.

In 1973 Hugh Porteous argued that Thom’s exact-unit hypothesis needed to compete with a pacing hypothesis. Builders taking a chosen number of paces could create recurring approximate sizes without maintaining an identical standard rod. He credited Thom’s painstaking work while challenging the inference drawn from the statistical test.⁵

Porteous illustrated the issue with 100 deliberately constructed circle diameters, clustered around ten paces of 33 inches each. These were invented numbers for a methodological example, not measurements of prehistoric monuments. We can reproduce a simpler part of his demonstration directly from his frequency table: divide each diameter by ten, calculate the spread of the resulting pace lengths, then calculate the uncertainty in their average.

Inputs: Porteous, p. 23, diameters 320–340 inches in two-inch steps, with frequencies 5, 5, 10, 10, 10, 20, 10, 10, 10, 5, 5. We use the sample standard deviation, dividing squared deviations by 99; the standard error is that deviation divided by √100. We are not reproducing his full historical significance test.

The average is ten times more precise than the spread of the individual paces. Collecting more observations can make an average impressively stable even when the underlying practice varies. Consequently, the error attached to an estimated common quantum cannot automatically be read as the precision with which every community reproduced a ruler.

Model flexibility matters too. In a 1977 exchange about Avebury, P. R. Freeman obtained different geometry by fitting arcs separately; Thom replied that individual positional errors and the additional freedom in those fits mattered. Freeman himself acknowledged fitting more parameters.⁶ That exchange concerns Avebury, not a direct reanalysis of Stonehenge. Its lesson for our test is to compare models with their assumptions and adjustable choices visible.

A fair modern investigation would specify the measurement endpoints, construction phases, candidate units, allowable error and alternative models before looking at the test sample. It would count unsuccessful fits and account for searching many units, rather than report only the best match. Searching dimensions, circumferences, subdivisions and stone subsets creates more opportunities for a pleasing answer. A low residual alone does not measure how surprising that answer is.

Could a chalk drum preserve a different measure?

There is a more tangible route into the question. Anne Teather, Andrew Chamberlain and Mike Parker Pearson proposed a link between Stonehenge’s dimensions and the chalk drums from Folkton and Lavant. Their alternative is a “long foot” of 0.3219 m. They suggest that winding cord around differently sized drums could reproduce approximately ten such feet: ten turns around Folkton III, nine around Lavant, eight around Folkton II or seven around Folkton I.⁷

This proposal brings artefacts into a debate otherwise dominated by monument plans. It also gives us something specific to check. Ten proposed long feet equal 3.219 m. Multiplying the paper’s published, rounded circumferences by the proposed turn counts gives:

These are our products of the authors’ Table 1 circumferences, not new measurements of the objects. Differences are relative to 3.219 m and rounded to two decimal places. The four results are within about 1.7% of that target; they are approximate matches, not four identical rope lengths.

The authors discuss a substantial qualification: chalk is poorly suited to repeated practical use, and they report no visible striations on the Lavant drum. They suggest that chalk objects might represent more durable wooden originals.⁷ That is an inference about function and lost objects, not a recovered measuring apparatus.

The long foot therefore deserves the same questions as the yard. Were circumference and turn count selected independently? How consistent are measurements taken at different heights on an imperfect cylinder? What new objects or structures would the proposed standard successfully predict? Tangible artefacts improve the kinds of evidence available, but their existence alone does not identify their use. Rejecting a universal megalithic yard need not mean rejecting measurement, practical geometry or every other proposed local standard.

The solar axis makes a different kind of claim

Stonehenge’s main axis connects the directions of midsummer sunrise and midwinter sunset. From the centre, the midsummer Sun rises close to the Heel Stone; looking the opposite way gives the midwinter setting direction. The orientation is expressed through the architecture and the approach, rather than recovered by choosing a length unit. English Heritage discusses both the solar alignment and the evidence that midwinter gatherings may have been particularly important.³

Direction and length are independent properties: enlarge a plan without rotating it and its axis still points the same way. The solar case can therefore be strong even if the yard remains unproven. Conversely, identifying a solar axis cannot calibrate a ruler.

Recent discoveries widen the setting. In June 2026 Wessex Archaeology announced a proposed solstitial alignment between two former timber posts, 120 m apart, at Bulford. The excavators place it around the time of Stonehenge’s first earthworks, roughly five centuries before the great sarsen settings, and describe it as a possible precursor.¹¹ This is an excavator announcement and an attributed interpretation. It adds a candidate for shared astronomical practice, not a test establishing Thom’s yard.

A full calendar is a further step. Timothy Darvill’s 2022 proposal reads the sarsen settings as a 365¼-day system: thirty days repeated over twelve months, five additional days and a four-year correction. Giulio Magli and Juan Antonio Belmonte challenge the numerical selection, the missing architectural expression of twelve and the astronomical argument. Darvill subsequently defended his interpretation.⁸⁹¹⁰ The controversy concerns how a functioning calendar would be demonstrated; it does not erase the solstitial orientation.

The Moon needs its own test as well. English Heritage describes a possible relationship between the Station Stone rectangle and the Moon’s extreme rising direction during the approximately 18.6-year standstill cycle, while explicitly treating deliberate lunar alignment as debatable.³ A solar orientation, a lunar interpretation, a calendar and a standard yard should not borrow one another’s confidence.

A connected Britain still needs evidence for its ruler

It would be a mistake to dismiss a shared measure simply because distant communities could not have communicated. The stones themselves make that picture of isolation difficult. In 2024 Anthony Clarke and colleagues compared mineral ages and chemistry in the Altar Stone with geological sources and supported an origin in the Orcadian Basin of northeast Scotland, at least 750 km from Stonehenge. They suggested sea transport; the exact quarry and journey were not identified.¹² This supersedes the Welsh attribution still present in some older accounts.

Such connections make the transmission of skills, designs or measuring conventions conceivable. They do not tell us whether people shared 0.829 m, 0.3219 m, a locally copied cord or several different practices. Geological provenance and metrological standardisation answer different questions. The stronger historical account allows the first to enrich the second without pretending to settle it.

What Stonehenge’s measures can establish

The monument rewards a more demanding kind of wonder. Its construction joins shaped stones, repeated geometry, changing arrangements and a relationship with the Sun. Thom’s survey remains valuable evidence even where his proposed design interpretation is disputed. The yard is testable precisely because it makes claims beyond the existence of a competent builder with a rope.

The most useful result of our calculation is that a close fit needs an explanation of the count, the measuring line and the independence of the evidence. One circle can suggest a ruler; it cannot uniquely name it. A successful shared-unit theory would carry a unit fixed in advance into new, well-defined measurements and outperform plausible alternatives with uncertainty included. Until then, the megalithic yard is an attributed hypothesis, not Stonehenge’s recovered standard.

Use the length converter to reproduce the site’s declared 0.829 m model, or read the megalithic-yard dossier for its numerical convention. For the next question, the Angkor Wat study shows why an exact proportion cannot independently reveal a module; the sacred-geometry framework extends the controls to other monuments.