A computer model for estimating travel speed based on topography, where velocity over a segment of trail is calculated as function of slope, could be derived using the function presented by Tobler (1993) for hiking and horseback riding (the source of cost paths shown in Figure 3-5). While these topography based calculations have serious limitations (Connolly and Lake 2006: 252-255), for general estimates over larger regions with measurable changes in terrain steepness these estimates are superior to the simple use of slope for estimating velocity. Such models would preferably be derived using original data from fieldwork, perhaps based on a contemporary study that takes into account the size and weight of the cargo animal, the amount and type of cargo, and the performance of the cargo animals based on trail conditions mapped using GPS receivers.

In the course of ethnoarchaeological fieldwork accompanying a llama caravan in Bolivia, Nielsen (2000: 449;2001: 184) notes that there were differences in the going and returning portions of a 2-3 month caravan journey that had implications for the overall travel speed. The out-going trip involved visits with companions and a variety of rituals at propitious locations along the route. During the return journey, in contrast, no rituals were performed but the animals moved more slowly because they were carrying large loads of produce uphill from the eastern lowlands; and they had been traveling for months and reportedly their feet hurt. Thus the lack of ritual performance and visiting and bartering saved time, but the walking speeds were slower, breaks were longer, and layover rest days were more frequent and longer.