William Oughtred

William Oughtred

Mathematical teaching and logarithmic instruments · 1574–1660 · England

Oughtred connects mathematical understanding with instrument design. A slide rule is useful because the positions of its marks encode a mathematical function. Learning to align its scales without understanding that function leaves important questions unanswered: where does the decimal point belong, how accurate is the result, and which operations are supported?

A careful historical account

Oughtred was baptized at Eton on 5 March 1574. That is not a documented birth date. He studied at King's College, Cambridge, became a clergyman, and served as rector of Albury from 1610 until his death in 1660. His Clavis Mathematicae appeared in 1631 and helped teach symbolic mathematics. His writing used the multiplication cross, but the history of mathematical notation cannot be reduced to a single inventor creating an entire symbolic language. MacTutor biography

Oughtred developed early slide-rule instruments, including circular forms. Priority requires qualification: Richard Delamain published Grammelogia in 1630, before Oughtred's Circles of Proportion appeared in English in 1632 through William Forster. The dispute also concerned teaching. Oughtred argued that students should understand the mathematical principles before operating the instruments. It is not justified to turn that position into an invented philosophy or a claim that he founded modern open education. MacTutor's discussion of the publications and dispute

A surviving 1654 slide rule made by Robert Bissaker provides material evidence of seventeenth-century instruments. It is Bissaker's artifact, not Oughtred's personal instrument, and cannot by itself settle the earlier priority dispute. Science Museum Group collection record

Worked mechanism: multiplication as a distance

The following is an original mathematical model of a straight slide rule. Assume two identical scales, each 250 millimeters long. On each scale, position p(x) for 1 ≤ x ≤ 10 is:

p(x) = 250 log10(x) millimeters.

The marks are not equally spaced. The distance from 1 to 2 is about 75.26 mm, whereas the distance from 8 to 9 is about 12.79 mm. The marks are arranged so that equal ratios occupy equal distances.

To multiply 2 by 3, place the movable scale's 1 at the fixed scale's 2. The movable scale's 3 is now:

p(2) + p(3)
= 250(log10(2) + log10(3))
= 250 log10(6)
≈ 194.54 mm

from the fixed origin. It aligns with 6. The hardware adds lengths; the encoding makes that addition represent multiplication.

What about 6 × 4 = 24? The total position exceeds one scale length. Align the movable scale's right-hand 10 with the fixed scale's 6 instead. The movable 4 then aligns with 2.4 because the displacement subtracts one length, corresponding to subtraction of 1 from the logarithm. The operator restores the power of ten using a separate magnitude estimate. A basic scale gives a significand, not the full answer's order of magnitude.

For 37 × 240, multiply the significands 3.7 × 2.4 = 8.88. Since 37=3.7×10 and 240=2.4×100, the result is 8.88×1000=8880. A rough check, 40×240=9600, makes an answer of 888 or 88800 implausible.

Accuracy comes from geometry

Suppose the combined positioning and reading error is at most 0.25 mm. This is a bound on the final displacement, not a claim that every individual alignment is that accurate. Dividing by the scale length gives a logarithmic error of at most 0.25/250=0.001.

The reconstructed answer therefore has a multiplicative error factor between 10^-0.001 and 10^0.001, roughly a 0.231% maximum relative deviation. A longer scale improves the bound if the physical reading error stays fixed. It does not make the mathematical result exact, and extra printed digits cannot overcome a poorly aligned instrument.

This also explains approximately uniform relative precision. Near a larger number, a fixed error corresponds to a larger absolute numerical error. For results near 10 and 1000, the same 0.231% allowance means about 0.0231 and 2.31 respectively. Relative error and absolute error answer different questions.

What transfers to computing

An interface often hides a representation. A good operator needs to know what the representation preserves and what remains outside it. Here, the scales preserve ratios while leaving signs and decimal placement to the user. Modern numerical software similarly requires units, domains, precision requirements, and independent reasonableness checks. This is a conceptual comparison, not evidence that all later interfaces descended directly from Oughtred.

Exercises

1. Compute 0.072 × 3500 using significands and powers of ten. Explain how a slide rule could give a plausible-looking but tenfold incorrect answer.
2. A 500 mm scale has a combined displacement error bounded by 0.20 mm. Derive the multiplicative error bound. Compare it with a 250 mm scale having the same displacement error. Do not simply say the longer one is "more accurate."
3. A museum has a slide rule dated 1654, and a book describing a circular slide rule was published in 1630. A curator concludes that the 1654 maker invented the slide rule in 1622. Identify what each item actually supports and what additional evidence the conclusion requires.

Compare your reasoning with the worked solutions

1. Write 0.072=7.2×10^-2 and 3500=3.5×10^3. The significand product is 25.2, so the full product is 25.2×10=252, or 2.52×10^2 after normalization. The scale may show 2.52 after wraparound. Interpreting that as 25.2 or 2520 is a decimal-placement error, not evidence of incorrect logarithmic marks. Estimate 0.07×3500=245 to select the correct magnitude.
2. For 500 mm, the logarithmic error is at most 0.0004, giving a ratio in [10^-0.0004,10^0.0004], with upper relative error about 0.09215%. For 250 mm it is at most 0.0008, giving about 0.18438%. The exponent bound halves; the exact relative bound is exponential, so describing it as exactly halved would be slightly inaccurate.
3. The artifact establishes that this instrument existed at its attributed date, subject to the catalog's dating evidence. The book establishes a published description by 1630. Neither proves who first conceived the device in 1622. That conclusion would need earlier dated documentation and an account of what qualifies as the same invention. Distinguish conception, construction, publication, and the surviving object's manufacture.