For any natural $n$, let $U_1,\dots,U_n$ be independent identically distributed random variables each uniformly distributed on the interval $[0,1]$. As usual, let $U_{n:1}\le\cdots\le U_{n:n}$ denote the corresponding order statistics. Consider
\begin{equation}\label{eq:G_i} G_i:=U_{n:i}-U_{n:i-1} \quad \text{for}\quad i=1,\dots,n+1, \end{equation} where $U_{n:0}:=0$ and $U_{n+1:n+1}:=1$.

One may refer to the $G_i$'s as the gaps or, as it is usually done in the literature, the spacings between the consecutive order statistics. Let now $$ G_{n+1:1}\le\cdots\le G_{n+1:n+1} $$ denote the ordered gaps $G_1,\dots,G_{n+1}$, so that the random sets $\{G_{n+1:1},\dots, G_{n+1:n+1}\}$ and $\{G_1,\dots,G_{n+1}\}$ are the same.

Fisher obtained the distribution of the largest gap, $G_{n+1:n+1}$. Fisher's result was generalized in this note, where the distribution of the $k$th smallest gap, $G_{n+1:k}$, was obtained, for each $k=1,\dots,n+1$.

Does anyone know other references to these or other related results in the literature?


1 Answer 1


I have just found out that this result is due to Irwin.

  • $\begingroup$ jstor.org/stable/pdf/2983737.pdf too? $\endgroup$ Nov 26, 2019 at 9:09
  • $\begingroup$ @BrendanMcKay : Thank you for your comment. Yes, that paper by Barton and David is where I found the reference to Irwin. $\endgroup$ Nov 26, 2019 at 14:42
  • $\begingroup$ Note also that exercise 667 in Whitworth's DDC Exercises precedes Fisher by roughly 30 years. $\endgroup$
    – esg
    Nov 27, 2019 at 18:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.