Recent writings on khipus (Andean knotted-cord records) invoke "writing wit
hout words," a near-synonym of Gelb's "semasiography," to argue that some A
merican media refer directly to cultural "things" without functioning as a
secondary code for speech. Sampson suggests that in principle such a system
could constitute a nonverbal "parallel language." However, no ethnography
actually shows whether Andean codes do so, much less reconstructs lost ones
. This study concerns a Peruvian village which inscribes its staffs of offi
ce in a code "without words." Fine-grained ethnography over several inscrip
tive cycles shows that staff code does function as a "parallel language." I
n doing so, however, it deviates interestingly from Sampson's model, for it
functions not to provide speech with a "direct reference" complement but t
o detach some areas of practice from the realm of discourse altogether. Con
sidered politically, this seemingly exotic method makes sense. Whether one
calls it "writing" depends on theoretical commitments in grammatology. High
ly inclusivist theories bear further development toward a more omnidirectio
nal ethnography of inscription.