On the other hand, a study to replicate the work of Amir, Stafford, Freshman, & Foa (1998, as cited in Gray & Lombardo, 2001) and Foa et al. (1995) by Gray & Lombardo achieved results entirely opposite the expectation that initial trauma narratives would be shorter, less descriptive, and contain less emotional content. Instead, their sample of college-enrolled participants with PTSD, who composed and submitted their trauma narratives almost^{Endnote 21} anonymously, tended to write the longest, most fully developed narratives about their traumas—longer than trauma, unpleasant, or pleasant narratives written by participants without PTSD. Gray & Lombardo suggest that perhaps the early, fragmented nature of trauma memory communications Foa and her colleagues report stems not from the organization of the memory itself, but rather from the mode of communication. Perhaps, they argue, early fragmentation or poor articulation of the trauma narrative relates to the embarrassment of sharing intimate details in oral disclosures to an unfamiliar (and not proven safe) therapist and may indicate not so much a poorly formed memory, but “avoidance of a trauma or of its meaning to the victim” (p. S182).