While RDF provides a powerful means to store knowledge, it can be cumbersome to represent and query collections of statements in context. To this end, we introduce a new a higher-level object, the Snippet, to hold a fragment of RDF that is about a single subject and made within a particular context. Each snippet may be represented in a standard form as a bag of reified statements or in a non-standard compact form using many fewer triples. Just as a quad serves as the internal representation for a single reified statement; so a compacted snippet serves as the internal representation for a bag of reified statements made about a single subject. Basic operations, such as storing and querying, may be performed on compact snippets without needing to expand them. Provided the application is aware of snippets this is possible using existing triple stores and query languages. In addition, the compacted snippets may always be expanded to permit reasoning using standard tools. Particular consideration is given to the use of snippets in distributed applications. Algorithms are presented for computing content-based identifiers and for handling security using split-capabilities. An implementation is described. 20 Pages