change of base theorem

Change of Base Theorem Made Easy: Formula, Proof, and Examples

By Mark Brown
AD

If you have ever stared at a scientific calculator that offers only two logarithm buttons — one labeled log for base 10 and one labeled ln for base e — while your textbook calmly asks for the logarithm of 250 in base 7, you have already run into the exact problem the change of base theorem was designed to solve. Logarithms are defined for infinitely many bases, yet the tools we carry around every day recognize only a couple of them. That gap between mathematical definition and practical computation is where this theorem earns its keep.

The change of base theorem is one of those quiet, unglamorous results that quietly powers a surprising amount of mathematics. It shows up when you evaluate an unfamiliar logarithm, when you compare algorithmic growth rates, when you rewrite derivatives in a friendlier form, and when you convert between units of information. Once you understand why the formula works, not just what it says, you stop memorizing it and start seeing it as an obvious consequence of how exponents behave.

Understanding the Change of Base Theorem

At its core, the theorem says that any logarithm can be rewritten as a ratio of two logarithms taken in a base of your choosing. In symbols, for a positive number x, a valid old base b, and a valid new base c, the identity reads log base b of x equals log base c of x divided by log base c of b. The old base disappears from the numerator entirely and reappears in the denominator, which is the detail students most often scramble.

AD
Change of basis | Chapter 13, Essence of linear algebra
Change of basis | Chapter 13, Essence of linear algebra
Reflection
Reflection
an image of some type of text with numbers and letters on it, including the letter b
an image of some type of text with numbers and letters on it, including the letter b

The reason this is not just an algebraic trick is that logarithms in different bases are proportional to one another. The graph of log base 2 of x and the graph of log base 10 of x have the same shape; one is simply a vertically stretched or compressed copy of the other. The change of base theorem identifies that stretch factor precisely as the reciprocal of the new base's logarithm, which is why the ratio is constant across all values of x.

The Formula, Written Out Clearly

Stated carefully, the theorem requires that x be strictly positive, that both bases be positive, and that neither base equal 1. Those conditions are not decoration. A base of 1 would make the logarithm meaningless because 1 raised to any power is still 1, and the denominator log base c of b would collapse to zero, producing division by zero.

a square with two sides and one side that has the same number of signs on it
a square with two sides and one side that has the same number of signs on it
a diagram with two intersecting sections and the same area for each section to be colored
a diagram with two intersecting sections and the same area for each section to be colored
Transformations
Transformations

A useful companion identity falls out immediately: log base b of x equals the reciprocal of log base x of b. This is just the main formula with the roles of the number and the base exchanged, and it is handy when the logarithm you want to compute has a large base and a small argument. For instance, log base 1000 of 10 is the reciprocal of log base 10 of 1000, which is 1 divided by 3, or roughly 0.3333.

A Short Proof Using Exponent Rules

Let y equal log base b of x. By the definition of a logarithm, that statement is equivalent to saying b raised to the power y equals x. Now apply log base c to both sides, an operation that is legitimate because x is positive and c is a valid base. The left side becomes log base c of b to the y, and the exponent rule for logarithms lets you pull the y out front, giving y times log base c of b.

AD

You are left with y times log base c of b equals log base c of x. Since b is not 1 and c is a valid base, log base c of b is nonzero, so dividing both sides is safe. The result is y equals log base c of x divided by log base c of b, which is exactly the theorem. Every step relies on nothing more exotic than the definition of a logarithm and the power rule, which is a good sign that the identity is fundamental rather than accidental.

two different signs that show the same sign and what they mean to be in each other
two different signs that show the same sign and what they mean to be in each other
🧐How do we understand a^3-b^3 better🥳?
🧐How do we understand a^3-b^3 better🥳?
Change of Base Formula
Change of Base Formula
two triangles are shown with arrows pointing to each other and one has the same side
two triangles are shown with arrows pointing to each other and one has the same side
The Unbreakable Bayes' Theorem | Towards Data Science
The Unbreakable Bayes' Theorem | Towards Data Science
an exercise sheet with two numbers and one number in the middle, which is written as 2
an exercise sheet with two numbers and one number in the middle, which is written as 2

Putting the Theorem to Work

Computation is the most visible use. Most calculators and programming languages implement natural logarithms and base 10 logarithms, so converting an awkward base into one of those two is the standard path to a numerical answer. But the theorem does more than produce decimals; it lets you restructure expressions so that cancellations become visible and comparisons become meaningful.

There is also a conceptual payoff. When physicists or computer scientists talk about "the logarithm" without specifying a base, they often can, because the change of base theorem guarantees that changing the base only multiplies the function by a constant. In asymptotic analysis, constant factors are frequently irrelevant, so log base 2 of n and log base 10 of n belong to the same growth class even though their numerical values differ.

AD

Evaluating Awkward Logarithms on a Calculator

Suppose you need log base 5 of 100. Your calculator has no base 5 key, so you rewrite it as log base 10 of 100 divided by log base 10 of 5. The numerator is exactly 2, because 10 squared is 100, and the denominator is approximately 0.69897. Dividing gives roughly 2.8614, which you can verify by checking that 5 raised to the 2.8614 power is close to 100.

The Relationship Between Geometry and Algebra
The Relationship Between Geometry and Algebra
geometry matters
geometry matters
Visual Proof of (a − b)² = a² − 2ab + b² Using Area Model
Visual Proof of (a − b)² = a² − 2ab + b² Using Area Model
a computer screen showing an image of a rectangle with two sides and one side that has
a computer screen showing an image of a rectangle with two sides and one side that has

The same technique works with natural logarithms and often produces cleaner intermediate numbers. For log base 2 of 7, write ln 7 divided by ln 2. The numerator is about 1.94591 and the denominator about 0.69315, so the answer is approximately 2.80735. Notice that you are free to choose either base 10 or base e; the quotient is identical either way, which is a practical demonstration that the theorem is consistent.

Simplifying Expressions and Comparing Rates of Growth

Consider the expression log base 4 of 8. Writing it as log base 2 of 8 divided by log base 2 of 4 turns a mystery into arithmetic: 3 divided by 2, or 1.5. Similarly, log base 9 of 27 becomes log base 3 of 27 divided by log base 3 of 9, which is 3 divided by 2. Choosing the new base to match the existing powers converts messy decimals into exact fractions.

AD

The comparison of growth rates uses the same idea. Because log base 10 of n equals log base 2 of n multiplied by log base 10 of 2, and that multiplier is a fixed constant near 0.30103, the two functions never diverge in order of magnitude. Algorithm analysts exploit this constantly: they state running times as O(log n) without fussing over the base, knowing the theorem guarantees the choice cannot change the asymptotic conclusion.

Nuances, Pitfalls, and Extensions

Like any compact identity, the change of base theorem invites a handful of predictable errors. The most common is dropping the denominator, writing log base b of x as log base c of x when the intended meaning was a ratio. The second most common is flipping the fraction, putting the old base's logarithm in the numerator and the number's logarithm in the denominator, which produces the reciprocal of the correct value.

Domain restrictions deserve equal attention. The theorem as stated applies only when x is positive and both bases are positive and different from 1. Attempting to apply it to a negative argument, or to a base of 1, produces a statement that looks algebraic but is simply undefined. When an expression involves variables, the careful move is to state the domain conditions alongside the rewritten form.

Mistakes That Cost Points

A frequent slip is confusing the argument with the base. Students sometimes write log base 5 of 100 as log 5 divided by log 100, which is backwards. The correct version puts the argument's logarithm on top and the base's logarithm on the bottom, so log 100 divided by log 5 is the right arrangement. Slowing down to label each piece before substituting prevents this entirely.

Another trap is applying the theorem to a product or sum inside the logarithm as though it were a separate step. The identity transforms the base, not the argument. If you have log base 3 of the quantity x plus 2, the change of base gives ln of x plus 2 divided by ln 3, and no further splitting is permitted. Keeping the argument intact is essential.

Where the Theorem Shows Up in Calculus and Computer Science

In calculus, the derivative of log base b of x is 1 divided by the quantity x times ln b. That formula exists only because the change of base theorem lets you rewrite the unfamiliar-base logarithm as ln x divided by ln b, after which the constant factor simply rides along through differentiation. Integration follows the same logic, converting bases before antiderivatives are taken.

Computer science embraces the theorem even more enthusiastically. Information theorists measure entropy in bits using base 2 logarithms, while statisticians prefer natural logarithms for likelihood calculations; the conversion between the two is a single constant. Whenever you see a logarithm in a complexity class, an information measure, or a decibel calculation, the change of base theorem is quietly standing behind the scenes making the arithmetic possible.

Keep the theorem in your mental toolkit and it will keep repaying you: next time a logarithm appears in a strange base, whether in a homework problem, a piece of code, or a research paper, you will know that the base is never really an obstacle, only a label you can swap at will.

AD