Interest in the inference of evolutionary networks relating species or populations has grown with the increasing recognition of the importance of hybridization, gene flow and admixture, and the availability of large-scale genomic data. However, what network features may be validly inferred from various data types under different models remains poorly understood. Previous work has largely focused on level-1 networks, in which reticulation events are well separated, and on a general network's tree of blobs, the tree obtained by contracting every blob to a node. An open question is the identifiability of the topology of a blob of unknown level. We consider the identifiability of the circular order in which subnetworks attach to a blob, first proving that this order is well-defined for outer-labeled planar blobs. For this class of blobs, we show that the circular order information from 4-taxon subnetworks identifies the full circular order of the blob. Similarly, the circular order from 3-taxon rooted subnetworks identifies the full circular order of a rooted blob. We then show that subnetwork circular information is identifiable from certain data types and evolutionary models. This provides a general positive result for high-level networks, on the identifiability of the ordering in which taxon blocks attach to blobs in outer-labeled planar networks. Finally, we give examples of blobs with different internal structures which cannot be distinguished under many models and data types.