The logarithm is defined as the operation that undoes exponentiation: for a>0,a=1, y=logax⟺ay=x (for x>0) — the logarithmic form and exponential form of the same fact, convertible into each other directly by matching base-to-base and exponent-to-log-value. The domain of logax is (0,∞) and its range is all of R; for the natural base e, logex is written lnx.
Three laws let logarithms convert products, quotients and powers into sums, differences and multiples: loga(bc)=logab+logac (product rule), logacb=logab−logac (quotient rule), and logabc=clogab (power rule) — repeatedly misapplying these (e.g. writing log(x+y)=logx+logy, which is false) is the single most common logarithm error. A logarithm in one base converts to any other base b via the change of base formula logax=logbalogbx; a special case is the reciprocal identity logax=logxa1.
Four logarithmic-inequality rules describe order: for base a>1, bigger argument means bigger log; for base 0<a<1, the order flips; and a log is positive exactly when its base and argument sit on the same side of 1, negative when they sit on opposite sides. For a base-10 (common) logarithm, splitting log10x=[log10x]+{log10x} into its integer part (the characteristic) and fractional part (the mantissa) gives a shortcut for counting digits: a positive integer with m digits has characteristic exactly m−1.