A scaling law describes how one quantity changes when another is multiplied by a factor. If $y\propto x^n$, then changing $x$ by a factor $a$ changes $y$ by $a^n$. Scaling often reveals the dominant physics before an exact calculation is possible.

For geometrically similar objects with characteristic length $L$, area scales as $L^2$ while volume and mass scale as $L^3$. If an animal were enlarged by a factor of $2$ without changing shape or density, its mass would increase by $2^3=8$, while a supporting cross-sectional area would increase only by $2^2=4$. The load per unit area would therefore double. This mismatch between surface and volume scaling explains many size effects in mechanics, biology, heat transfer, and transport.

An order-of-magnitude estimate keeps only enough precision to identify the relevant scale. Quantities are rounded to convenient powers of ten and the result is interpreted logarithmically rather than as an exact prediction. For example, estimate the kinetic energy of a $1000,\mathrm{kg}$ car moving at about $30,\mathrm{m/s}$: $$K\sim \frac12(10^3)(3\times10^1)^2\approx 5\times10^5\ \mathrm{J}.$$ The useful conclusion is that the energy is of order $10^6,\mathrm{J}$, not that the estimate is accurate to the last digit.

Scaling and estimation are especially useful for checking calculations. A result that predicts a nanosecond for a planetary orbit or a gigajoule for lifting a book is suspect even if every algebraic step appears legal. Exact models refine physical reasoning; they do not replace the habit of asking what dependence and magnitude should be plausible.