# How signed integers are represented using 16 bits Source: http://www.xkcd.com/571/

This probably requires a little explanation with the nuances of how integers are stored in computers. A 16-bit signed integer simulates binary digits with 16 digits, where the first digit represents either positive ( $0$) or negative ( $1$). Because this first digit represents positive or negative, the counting system is a little different than regular counting in binary.

For starters, $0000000000000000$ represents $0$ $0000000000000001$ represents $1$ $0000000000000010$ represents $2$ $\vdots$ $0111111111111111111$ represents $2^{16}-1 = 32767$

For the next number, $1000000000000000$, there’s a catch. The first $1$ means that this should represent a negative number. However, there’s no need for this to stand for $-0$, since we already have a representation for $0$. So, to prevent representing the same number twice, we’ll say that this number represents $0 - 32768 = -32768$, and we’ll follow this rule for all representations starting with $1$. So $0000000000000000$ represents $0-32768 = -32768$ $0000000000000001$ represents $1 - 32768 = -32767$ $0000000000000010$ represents $2 - 32768 = -32766$ $\vdots$ $0111111111111111111$ represents $32767-32768 = -1$

Because of this nuance, the following C computer program will result in the unexpected answer of $-37268$ (symbolized by the sheep going backwards in the comic strip).

main()

{

short x = 32767;

printf(“%d \n”, x + 1);

}

For more details, see http://en.wikipedia.org/wiki/Integer_%28computer_science%29

This site uses Akismet to reduce spam. Learn how your comment data is processed.