The so-called prisoners’ problem, in which two individuals attempt to communicate covertly without alerting a “warden” who controls the communications channel, has taken a number of forms, adorned with various assumptions or requirements which make the problem more or less difficult. One assumption which makes the problem considerably more managable is that the participants are allowed to share some secret information (such as an encryption key) prior to imprisonment. Another assumption, which makes the problem much more difficult, is that the warden be allowed to modify messages sent between the prisoners as well as read them. This paper describes techniques for pure steganography, in which no secret information needs to be shared before imprisonment. First, a modification of an existing protocol will be shown to admit pure steganography if the warden is not allowed to modify the contents of the channel. Then, a technique will be described that allows pure steganography between two prisoners in the presence in the presence of an active (content-modifying) warden.