This paper unpicks the assumption that because social networks underpin social capital, they directly create it – more of one inevitably making more of the other. If it were that simple, the sheer quantity of networks criss-crossing a defined urban space would be a proxy measure for the local stock of social capital. Of course the interrelationships are more complex. Two kinds of complication stand out. The first is specific: networks have both quantitative and qualitative dimensions, but the two elements have no necessary bearing on each other. The shape and extent of a network says nothing about the content of the links between its nodes. Certainly the line we draw between any two of them indicates contact and potential connection, but what kind of contact, how often, how trusting, in what circumstances, to what end…? Reliable answers to these questions need more than surface maps or bird’s eye accounts of who goes where, who speaks to whom. The second complication is a general, not to say universal, difficulty. We are stuck with the fact that sociological concepts – networks, social capital and trust included – are ‘only’ abstractions. They are ways of thinking about the apparent chaos of people behaving all over the place – here, to make it worse, in multi-cultural urban environments – but none of them is visible to be measured, weighed or quantified. This does not make the concepts ‘untrue’, and it should not stop them being useful. My hope is that we can find a nuanced perspective which will at least make the complications intelligible. At best, a multi-layered model will account for diversity in the nature of trust; and for variations in the way social capital is hoarded or distributed within and across ethnic boundaries. It would be contribution enough if we were able to specify the conditions which cause social capital, as Puttnam formulates it, to be exclusionary or inclusionary in its effect..