Every logarithm is an exponent
A logarithm is a power. is the power to which a must be raised to give x, so every fact about logarithms is a fact about exponents, read backward. The laws of logarithms come from the laws of exponents in exactly this way.
The law of exponents behind the first law of logarithms is about multiplying powers of the same base. means (2 × 2 × 2) × (2 × 2 × 2 × 2), which is seven 2s multiplied: . Multiplying powers of one base adds the exponents, 3 + 4 = 7.
The product law
Read that fact in logarithms to base 2. , and . So , and , and 3 + 4 = 7. The logarithm of the product is the sum of the logarithms.
The same argument works for any positive numbers. Take the common logarithm, to base 10, and write a and b as powers of 10: a = 10ᵐ and , so that log a = m and log b = n. Then , and so log(ab) = m + n = log a + log b.
For example, log 2 = 0.30103 and log 3 = 0.47712, to 5 decimal places, and log 6 = log(2 × 3) = 0.30103 + 0.47712 = 0.77815, which is what a calculator gives for log 6. The argument used base 10, but any base gives the same law, as long as every logarithm in it has the same base.
The top row is x, the powers of 2 from 1 to 128, and the row under it is . Multiplying two numbers in the top row adds their logarithms: 8 × 16 = 128, and 3 + 4 = 7.
The quotient law
Dividing powers of one base subtracts the exponents: . With a = 10ᵐ and again, , so . Dividing inside the logarithm becomes subtracting outside it.
In base 2: 128 ÷ 8 = 16, and . In base 10: .
The power law
A power is repeated multiplying, so the product law gives the power law at once for a whole number n. . In general, : the exponent comes to the front as a multiplier.
The law of exponents behind it is the power of a power, which multiplies the exponents. Write x = 10ᵐ, so log x = m. Then , and mn = n log x. This argument does not need n to be a whole number, so the power law holds for every n. For n = ½ it says , and for n = −1 it says .
For example, , and .
Two values that are the same in every base
for every base, so . , so . These fit the laws: log(x × 1) = log x + log 1 = log x + 0, as it must be.
The laws together
The laws can combine several logarithms into one. For 2 log 3 + log 4 − log 6, first use the power law, . Then the product law gives log 9 + log 4 = log 36, and the quotient law gives .
They can also split one logarithm into several. .
What the laws do not say
The laws turn multiplying into adding. They say nothing about the logarithm of a sum, and log(a + b) is not log a + log b. Test it with numbers: log(10 + 10) = log 20 = 1.3010, but log 10 + log 10 = 1 + 1 = 2, which is log 100, the logarithm of 10 × 10.
Nor is the product of two logarithms the logarithm of the product. log 100 × log 10 = 2 × 1 = 2, but log(100 × 10) = log 1000 = 3. And a quotient of logarithms is not the logarithm of the quotient: , while .
The power law brings the exponent of x to the front. It does not apply to a power of the whole logarithm: means log x × log x, which is not 2 log x. At x = 1000, , while 2 log 1000 = 2 × 3 = 6.
Three pairs of numbers a and b: 2 and 5, 10 and 10, 4 and 25, with common logarithms to 4 decimal places. The first two columns agree every time; the third, the logarithm of the sum, never does.
The usual mistakes
Adding inside the logarithm. log 4 + log 5 is log(4 × 5) = log 20, not log(4 + 5) = log 9.
Subtracting inside the logarithm. log 20 − log 4 is , not log 16.
Cubing the logarithm. is 3 log x, not : the power sits on the x inside, and it comes down as a multiplier.
Mixing bases. , but is about 6.32 and is about 1.90, and neither is 4. The laws hold only when every logarithm has the same base.
A power law as a straight line
The next problem fits a model of the form to measured data. Take the logarithm of both sides and use two laws at once. The product law splits the right side: . The power law brings n to the front: log B = log a + n log M.
Write X for log M and Y for log B. The equation becomes Y = nX + log a, which is a straight line, with gradient n and intercept log a. So if the points with coordinates log M and log B lie on a straight line, the data follow a power law: the gradient of the line is the power n, and the intercept is log a, so a is 10 to the power of the intercept.
Worked example: The Resting Energy Use of Mammals Against Their Mass: A Power Law on Log-Log Axes, and a Prediction for a Bear
Question A biologist's table gives the mass M kg and the resting energy use B kilocalories per day of four mammals. For M = 1, 16, 81, 625 the values of B are 70, 560, 1890, 8750. To 3 decimal places, the values of log10 M are 0, 1.204, 1.908, 2.796 and the values of log10 B are 1.845, 2.748, 3.276, 3.942. (a) Show that the points (log10 M, log10 B) lie on a straight line, and find its gradient and its intercept. (b) The biologist fits the model B = a × Mn. Find n and a, and use the model to predict the resting energy use of a bear of mass 256 kg, to 3 significant figures.
1.Take logarithms to base 10 of B = a × Mn. The logarithm of a product is a sum, and the power law brings n down: log10 B = log10 a + n log10 M. This has the form Y = c + mX, with X = log10 M and Y = log10 B.
Taking logarithms of B = a × Mn gives log10 B = log10 a + n log10 M, a straight line in log10 M and log10 B. 2.From (0, 1.845) to each of the other points the gradient is the same: 2.748 − 1.8451.204 = 0.9031.204 = 0.75, 3.276 − 1.8451.908 = 1.4311.908 = 0.75 and 3.942 − 1.8452.796 = 2.0972.796 = 0.75. So the four points lie on one straight line.
The line rises 2.097 while it runs 2.796, so its gradient is 0.75, and the other points give the same gradient. 3.(a) The gradient is 0.75. The mammal of mass 1 kg has log10 M = 0, so its point (0, 1.845) is on the vertical axis, and the intercept is 1.845.
(a) The gradient is 0.75, and the line meets the vertical axis at 1.845, the point of the 1 kg mammal. 4.The gradient is the power itself, so n = 0.75. The intercept is log10 a, so a = 101.845 = 69.98, which is 70 to 2 significant figures: the resting energy use of a mammal of 1 kg. The model is B = 70 × M0.75.
The gradient is n itself, so n = 0.75. The intercept is log10 a, so a = 101.845 ≈ 70. 5.(b) For the bear, 256 = 44, so 2560.25 = 4 and 2560.75 = 43 = 64. Then B = 70 × 64 = 4480 kilocalories per day. Check on the line: log10 256 = 2.408 and 1.845 + 0.75 × 2.408 = 3.651, and 103.651 ≈ 4480.
(b) B = 70 × M0.75, and for the bear of 256 kg, B = 70 × 64 = 4480 kilocalories per day.
Answer: (a) the gradient from the first point to each of the others is 0.75, so the points lie on a straight line with gradient 0.75 and intercept 1.845; (b) n = 0.75 and a = 70, so B = 70 × M0.75; about 4480 kilocalories per day
Common mistakes
- Turning the gradient back with a power of 10, as n = 100.75 = 5.62. On log-log axes n multiplies log M directly, so the gradient is n itself. Only the intercept, log10 a, has to be turned back into a.
- Plotting log10 B against M instead of against log10 M. That plot straightens B = a × bM, where M is an index. In a power law M is the base, so its logarithm must go on the horizontal axis.