Langmuir (2004, 185–99) dedicates a chapter of his book to river crossings, which indicates that this is by no means a trivial matter even in modern times. He points out the dangers even small streams can pose when fed by torrential rain or melting snow. He comes to the conclusion that fording a river by whatever method always involves some degree of risk. Consequently, he recommends checking the alternatives thoroughly and to adopt fording only when the alternatives to crossing are more hazardous than the crossing itself. The following criteria are given by Langmuir to select a crossing point: (a) a stream can be crossed more easily where it is braided into a number of smaller streams; (b) a river running into a lake normally flattens out and slows down, so fording a river near its mouth is often possible; (c) the area selected should be free of obstructions such as fallen trees or boulders (submerged or otherwise) and the river bed as even as possible; (d) the outside of bends should be avoided, where the current is usually strongest. Nicke (2001, 15) describes the criteria for an adequate ford location in the Bergisches Land, Germany: the stream must be fast-flowing at the ford position so that no sandbanks are formed. The stream should be not too deep and crossing at meanders should be avoided as the riverbed might be deeper in some of these areas. In addition, the route should leave the muddy areas near the stream as soon as possible, climbing to the ridgeway again. The automatic identification of suitable ford locations according to the criteria of Langmuir or Nicke requires detailed geographical data, and the landscape might have changed so that the locations found might not have been fordable in the past.
Fábrega Álvarez and Parcero Oubiña (2007) note for their study area in Spain that the crossing points of the rivers present crucial points in the landscape to articulate movement on the whole. This is a fact for many landscapes, and therefore putting some effort into the location of old fords in historic sources, maps, and by analysing place names could well be a rewarding task.
Eichfeld (2005) uses a factor of 5.5 for small streams. Kondo (2008) suggests the factor 3 for travelling in a water zone. Van Leusen (2002, chapter 16, 13) increased the cost component for streams until the resulting LCPs did not cross the streams repeatedly, ending with a cost coefficient of 88 for a major stream, 22 for larger streams and 5.5 for smaller ones. With a correct application of Pandolf's formula (see Section 5.2), van Leusen probably would have found lower cost coefficients for the streams considered. For LCP reconstructions of historical routes in the Bergisches Land several multipliers were tested: for larger streams, a factor between 10 and 20 is appropriate, for small streams multiplier selection in the range of 5 to 10 created the most reasonable results (Herzog 2009b; 2010; 2013a).
Very small creeks may not have been considered as a significant barrier because people could jump over the creek. But even if this was possible, the wet surroundings might have prevented people from walking in the creek valley, at least during wet seasons. Nevertheless, a fresh water supply was needed both for travellers and for pack animals. It is for this reason that ancient tracks crossing small creeks, especially near the source, have been recorded (Nicke 2001, 19). According to Langmuir (2004, 238, 249), a walker in Britain requires about 2.5 litre of water on average, or to be more precise, 0.5 litre for each 1000 kcal expended in winter and twice as much in summer.