Steady-state infiltration rates were measured at three instrument scal
es within a 4.0 x 4.0-m field plot. The three scales were 4.0 x 4.0 m
(scale S(L), n = 1), 1.0 x 1.0 m (scale S(L/4), n = 16), and 0.25 x 0.
25 m (wale S(L/16), n = 256). After a 30-day ponding period under a co
nstant hydraulic head, infiltration measurements were made at each mea
surement scale. Even though the entire area was sampled at each instru
ment scale, the average infiltration rate decreased with decreasing si
ze of infiltrometer. The infiltration rate measurements were adjusted
assuming an ''apparent'' stagnation zone of 4 cm along the boundaries
of each infiltrometer. This adjustment produced average infiltration r
ates of 1.96 cm/h for each of the smaller measurement scales. This val
ue was in excellent agreement with the final value measured at scale S
(L) upon completion of the experiment (1.97 cm/h). However, the value
for the 'apparent' stagnation zone was found to be valid only in an av
erage sense, because the correlation between the adjusted average valu
e of the S(L/16) measurements and the adjusted value of the S(L/4) mea
surements was not significant. The 256 measurements at scale S(L/16) w
ere neither normal nor lognormally (natural log (ln)) distributed, alt
hough the latter. provided a somewhat better representation. Too few m
easurements were available to determine the probability distribution f
or scale S(L/4). The regularized semivariogram (range of 0.7 m) for sc
ale S(L/16) was deconvoluted to provide estimates of the point semivar
iograms for both the actual and adjusted measurement scales, leading t
o spatial ranges of 0.21 and 0.30 m for actual and adjusted, respectiv
ely. With or without adjustment, dispersion variance analyses illustra
ted that the spatial structure estimate obtained from measurement scal
e S(L/16) was inconsistent with that obtained from measurement scale S
(L/4). Thus, we conclude that the infiltrometer instrument fundamental
ly alters the infiltration process in such a way that measurements are
only meaningful in a relative sense.