1: \begin{abstract}

2: We introduce \textit{weaves}, which are random sets of non-crossing {\cadlag} paths that cover 

3: space-time $\ov{\R}\times\ov{\R}$. 

4: The Brownian web is one example of a weave,

5: but a key feature of our work is that we do 

6: not assume that particle motions have any particular distribution.

7: Rather, we present a general theory of the structure, characterization and weak convergence of weaves.

8: 

9: We show that the space of weaves has a particularly appealing geometry, 

10: involving a partition into equivalence classes under which 

11: each equivalence class contains a pair of distinguished objects known as a \textit{web} and a \textit{flow}. 

12: Webs are natural generalizations of the Brownian web 

13: and the 

14: flows provide a pathwise representations of stochastic flows. 

15: Moreover, there is a natural partial order on the space of weaves,

16: characterizing the efficiency with which paths cover space-time, 

17: under which webs are precisely minimal weaves

18: and flows are precisely maximal weaves. 

19: This structure %, whilst interesting in its own right,

20: is key to establishing weak convergence criteria for general weaves,

21: based on weak convergence of finite collections of particle motions.

22: \end{abstract}

23: