1: \begin{abstract}   

2: 	Consider a random uniform sample of $n$ points in 

3: 	a compact region $A$ of Euclidean 

4: 	$d$-space, $d \geq 2$, with a smooth 

5: 	or (when $d=2$) polygonal boundary.

6: 	Fix $k \in \NN$.

7: 	Let $T_{n,k}$  be the threshold $r$ at which the 

8: 	geometric graph on these $n$ vertices with distance

9: 	parameter $r$ becomes $k$-connected.

10: 	We show that if $d=2$ then $n (\pi/|A|) T_{n,1}^2 - \log n$

11: 	is asymptotically standard Gumbel.

12: 	For $(d,k) \neq (2,1)$,

13: 	it is $n (\theta_d/|A|) T_{n,k}^d -

14: 	(2-2/d) \log n - (4-2k-2/d) \log \log n$ that converges 

15: 	in distribution to a nondegenerate limit,

16: 	where $\theta_d$ is the volume of the unit ball. 

17: 	The limit

18: 	is Gumbel

19: 	with scale parameter 2 except when $(d,k)=(2,2)$ where

20: 	the limit is two component extreme value distributed.

21: 	The different cases reflect the fact that boundary effects are

22: 	more more important in some cases than others.

23: 	We also give similar results for the largest $k$-nearest

24: 	neighbour link $U_{n,k}$ in the sample, and show

25: 	$T_{n,k}=U_{n,k}$ with high probability.

26: 	We provide estimates on rates of convergence and

27: 	give similar results for Poisson samples in $A$.

28: 	Finally, we give similar results even for non-uniform

29: 	samples, with a less explicit sequence of centring

30: 	constants.

31: \\

32: \noindent{\bf Keywords:} Connectivity threshold; weak limit;

33: 	Poisson process; Gumbel distribution. \\

34: %\noindent{\bf AMS 2010 Classification:}

35: \end{abstract}

36: