There are opportunities to display information in different ways. Node-link methods are common for multivariate networks, but there are a lot of other possibilities. To inspire new visualization techniques we look for novel visual metaphors for multivariate networks and discuss the data visual mapping process involving them, we discuss where ideas come from, and consider the ideation process itself. The metaphors are organized into nature, non-physical, man-made, and visualization-inspired classes. Besides exemplar work from each class, we present a gallery of potential new metaphors.