I argue against preservationism, the epistemic claim that memories can at most preserve knowledge generated by other basic types of sources. I show how memories can and do generate knowledge that is irreducible to other basic sources of knowledge. In some epistemic contexts, memories are primary basic sources of knowledge; they can generate knowledge by themselves or with trivial assistance from other types of basic sources of knowledge. I outline an ontology of information transmission from events to memory as an (...) alternative to causal theories of memory. I derive from information theory a concept of reliability of memories as the ratio of retrieved information to transmitted information. I distinguish the generation of knowledge from reliable memories from its generation from unreliable memories. Reliable memories can generate new knowledge by forming together narratives and via colligation. Coherent, even unreliable, memories can generate knowledge if they are epistemically independent of each other and the prior probability of the knowledge they generate is sufficiently low or high. Ascertaining the epistemic independence of memories and eliminating possible confounders may be achieved through the generation of knowledge from independent memories in different minds, when memories are primary basic sources of knowledge and the testimonies that report them are trivial. (shrink)