Reprinted from SOIL SCIENCE
Vol. 92, No. 2
August, 1961
Copyright © 1961 by The Williams & Wilkins Co.
Printed in U.S.A.
VARIATION IN pH AND BUFFERING CAPACITY OF THE ORGANIC LAYER
OF GREY WOODED SOILS
H. VAN GROENEWOUD
Canada Department of Agriculture'
Received for publication December 13, 1960
During a study of the pH of forest soils in the
Candle Lake area of Saskatchewan, some
particular problems involving a wide variation
of pH and buffering capacity, were encountered.
This led to experiments on the effects of sample
size, variation of buffering capacity, and seasonal
pH fluctuations on the frequency distribution of
pH.
DESCRIPTION OF SOILS AND METHODS
The soils studied belong to the grey wooded
(podzolic) soils as described by Moss and St.
Arnaud (7).
The pH of the soil was determined in the field,
or within a few hours after sampling in the
laboratory, using a Beckman pH meter with a
combination glass electrode. The samples were
prepared as a soil paste according to Doughty's
method (4). In order to study the frequency
distribution of pH values in individual soil
layers, the acidity of 100 small and 100 large
samples (approximately 3 cc. and 60 cc., respectively), taken at random from the F, H, and A
soil layers within a 50 x 50 foot area, was determined at 20 different sites.
Titration curves were constructed from data
obtained by titrating samples of approximately
1 g. with 0.02 N HC1 and KOH. The results
were calculated on a basis of 10 g. oven-dry
(105°C.) weight.
Buffering in each soil layer was studied by
measuring the pH of individual 1-g. samples,
and remeasuring the pH after a thorough mixing
of different numbers (2 or 20) of these samples.
The measured pH of the mixed sample was then
compared with the calculated average. Data
converted to ion concentrations were used to
calculate the averages. Three soil layers at five
1 Contribution No. 703, Forest Biology Division,
Research Branch, Canada Department of Agriculture, Ottawa, and No. 66, Research Station,
Canada Department of Agriculture, Saskatoon,
Saskatchewan.
100
different sites were tested. The difference between
the measured and the calculated pH appears to
be a useful indicator in the study of buffering
for pH.
The pH of large composite samples (composed
of 20 samples of 60 cc. each, taken at random
within each sample plot from each soil layer)
was measured at different times during the
growing season. Each composite sample was
thoroughly mixed and left for a suitable time
(24 hours) to attain equilibrium before measuring
the pH.
RESULTS AND DISCUSSION
In several sites, a remarkably wide range of
pH (frequently 4-8.5) was found within each soil
layer when sampled at the same time. It was
even more significant that large differences in pH
occurred within areas as small as a few square
inches (table 1). This can result in variation in
measured pH, with changes in sample sizes.
Furthermore, buffering capacity can affect the
mean pH and the frequency distribution, as will
be shown later.
Samples collected close together showed wide
variation in behavior on titration with bases and
acids (fig. 1). These titration curves, however,
are considered to give only an indication of the
buffering capacity, as the latter can not very well
be derived in this way because of adsorption and
other phenomena. Among other things, chelation
reactions can occur (1), and the use of different
bases or acids can result in different titration
curves (8). The curves could not be used to
calculate the pH of mixed samples.
The experiments in which two individual 1-g.
samples were mixed also showed differences
between buffering capacities of small samples
from the same soil layer (table 2). The same was
true when 20 individual 1-g. samples were mixed
(table 3). In the tables, the differences between
measured and calculated averages are in pH units,
and, as they occur at different places in the pH
pH OF GREY WOODED SOILS range, they are not strictly comparable. With
this in mind an extra column was added to the
table to show the equivalent differences in
hydrogen ion concentration. In every instance
the measured pH was higher than the calculated
one. This indicates that the sample with the
higher pH value also had a higher buffering
capacity than the sample with the lower pH. If
the different buffering capacities were distributed
at random between higher and lower pH samples,
this would tend to decrease the differences
between measured and calculated pH in the
large composite samples.
It is evident from these experiments that not
only can there be a wide range of pH among
small samples of one soil layer, but that the
buffering capacity of the soil can be equally
variable. The use of different sample sizes in the
measurement of pH can result in quite different
range and form of distributions, if the buffering
TABLE 1
pH of pairs of samples taken less than 1 inch apart,
from F or H horizons
5.- and 6.4
4.2 and 5.8
4.5 and 5.2
4.6 and 5.6
5.- and 6.3
5.1 and 6.3
5.2 and 6.5
4.9 and 7.5
4.8 and 6.5
6.2 and 7.1
5.7 and 7.3
6.- and 7.4
5.4 and 6.6
5.2 and 6.2
5.2 and 6.5
6.2 and 7.4
6.1 and 7.8
4.4 and 6.2
Calculated with pH data converted to ion
concentrations.
101
capacity of the soil is variable. This effect is
shown in figures 2 and 3. Figure 2 represents the
theoretical frequency distribution of the pH of
a series of mixed samples determined by calculating the mean of the pH of the samples, assuming
uniform buffering capacity of the soil and
disregarding other phenomena. Figure 3 shows
the distribution of the same series but includes
the effects of differences in buffering capacity.
Comparison of the two distributions shows the
tendency of the heterogeneous buffering to
decrease the range of the pH and to produce a
distribution with a high concentration in one
class, with a slight shift of the range and the
mode towards a higher pH. This effect may lead
to misinterpretation; as a rule, sample sizes for
pH measurements are very indefinite and are
seldom mentioned in literature. Within a series,
individual samples which are not equal in size
can severely affect the distribution curve.
The plotting of the pH of small samples from
20 different sites and each of 3 soil horizons,
resulted in many different forms of frequency
distributions. There seems to be some dispute as
to which way pH data should be treated for
statistical analysis. Daubenmire (3) suggested
the transformation of pH data to hydrogen ion
concentrations or equivalent arithmetical values
before calculating the means. This is in order, as
pH values are logarithmic values and should be
treated as such. Shiue and Chinn (9) pointed out,
however, that the frequency distributions of
hydrogen ion concentrations show extreme
11
10
9
8
7
6
5
4 0
0
150
C.C. 1 50 N. HCL.
100
50
0
50
C.C.
100
150
1/50 N. KOH.
Fro. 1. Titration curves of two humus samples (H-horizon).
102
VAN GROENEWOUD
TABLE 2
Effect of differences in buffering capacity of single
samples on the pH of two samples mixed
Average of Mixed
Samples
pH of Samples*
(1)
(3)
(5)
(7)
(9)
(11)
(13)
(15)
(17)
(19)
5.0
4.8
4.9
4.6
4.1
4.6
4.4
4.6
4.4
4.5
(2)
(4)
(6)
(8)
(10)
(12)
(14)
(16)
(18)
(20)
6.3
6.0
6.9
6.6
6.1
6.1
6.0
5.8
5.9
6.1
Differences
Hydrogen ion
Calculatedt Measured pH units concentration
5.3
5.1
5.2
4.9
4.4
4.9
4.7
4.9
4.7
4.8
5.7
5.4
5.3
5.4
5.2
5.4
5.4
5.4
4.9
5.9
0.4
0.3
0.1
0.5
0.8
0.5
0.7
0.5
0.2
1.1
30
40
13
85
337
85
160
85
75
147
* Sample number in parentheses.
t Calculated with pH data converted to ion
concentrations.
g. ions/liter X 108.
skewness. As the frequency distributions which
they studied approached normal distributions
more closely than the corresponding distribution
of the hydrogen ion concentrations, they proposed
to retain pH values as such, without transformation, for statistical analysis other than
finding sample means. They only investigated,
however, the pH distributions of three soils.
In the present study several distributions were
found which departed far from normality. They
could not be regarded as approaching normal
distribution through the functioning of the
central limit theorem, as in the case of Shiue and
Chinn Both positively and negatively skewed
distributions were found. As a consequence,
transformation of data to normalize frequency
distributions cannot be employed, especially
where comparisons have to be made with small
numbers of measurements.
Before attaching too much value to the normality of the frequency distribution as affected
by sample size and buffering capacity, the effect
that seasonal pH fluctuations (2, 6) can have on
frequency distributions should be shown. Figures
4 and 5 show frequency distributions from the
same plot and the same soil layer in, respectively,
August, 1958, and May, 1959. It is evident that
the pH did not change to the same extent
throughout the soil. It seems that the less the
buffering for pH at a certain point in the soil,
the greater was the change in pH. Possibly the
pH increases again during the winter by diffusion
of certain ions. The remarkable similarity of pH
distribution curves of measurements of large
samples taken in the spring and the fall (figs. 6
and 7), plus the fact that sample size practically
had no effect on distribution curves in the spring
(figs. 5 and 6), are in support of the forementioned
thesis. The foregoing shows that any one distribution taken at a certain time of the year cannot be compared statistically with a distribution
from another site taken under different circumstances at another time; unless these changes
can be taken into account the distribution loses
much of its statistical value.
An attempt was made to find a method of
measuring pH which would give a value not
subject to seasonal fluctuations and which would
lend itself to comparison of pH for different sites.
The pH of large composite samples appeared to
solve the above-mentioned difficulties. When this
method was used, pH fluctuations within one
soil layer measured through one entire season
amounted to less than 0.2 pH unit. The mean pH,
measured on large composite samples, therefore,
constitutes the only statistic that is not subject
to much variation due to sample size, variation
in buffering capacity, or seasonal fluctuations.
It can be used directly in statistical analysis if
the values are normally distributed; if not, suitable transformations have to be used.
The "measured mean pH" does not, however,
supply information on the range of pH within a
forest site. To obtain this information, 100 measurements were taken at random within each
TABLE 3
Effect of differences in buffering on pH of mixed
samples (20 samples mixed)
Averages
Sample Series
1-20
21-40
41-60
Differences
Calculated*
Measured
pH units
Hydrogen
ion concentrationt
4.35
4.47
4.9
4.85
4.8
4.97
+0.50
+0.33
+0.07
+300
+160
+ 20
* Calculated with pH data converted to ion
concentrations.
t g. ions/liter X 107.
pH
103
OF GREY WOODED SOILS
30
a.
20
n/-n
IL
0
10
0
z
0
0 4.
4.5
5
5.5
pH
6
FIG. 2. Frequency distribution of calculated means.
70
60
50
20
10
0
0
5.5
5
4.5
6
pH
F IG. 3.
Frequency distribution of measured means.
20
15
L.L. 10
0
0
4
5
6
7
8
9
pH
FIG. 4.
Frequency distribution of the pH of 100 small samples collected in the fall of 1958.
sample plot on small samples of each soil horizon.
It is usually impossible to calculate the standard
error of each site, as has been done in some statistical analyses (5), because the "measured
mean pH" is located eccentrically in most of the
pH ranges of the soils.
In the analysis of the pH data of grey wooded
soils then, the maximum, the minimum, and the
"measured mean pH" have to be treated separately, which increases considerably the burden
of calculation.
SUMMARY
Sample size can be important in pH determinations, and care should be taken in the
choice of size for any particular study, as this
can affect the shape of pH distribution curves.
Heterogeneous buffering for pH in soils
can be an important factor which should be considered in choosing sample size.
Frequency distributions of pH are not
always normal or near normal.
Care should be taken to evaluate the fore-
▪
104
VAN GROENEWOUD
40
35
30 —
25
0
0
6
LL
20
15
10
5
4
7
6
8
9
pH
F IG. 5. Frequency distribution of the pH of 100 small samples collected in the spring of 1959 (same
site, same soil horizon as fig. 4).
50
40
0
30
•-^ 20r
0
0
7
5
4
8
pH
FIG. 6. Frequency distribution of the pH of 100 large samples collected in the spring of 1959 (same
site, same soil horizon as figs. 4 and 5).
40
0J
30
20
LL
0
0
3
sj
4
pH
F IG. 7. Frequency distribution of the pH of 100 large samples collected in the fall of 1958 (same site,
same soil horizon as figs. 4, 5, and 6).
mentioned factors so that they can be taken into
account in the statistical analysis wherever this
is deemed necessary.
(e) It is suggested that the measured pH of
large composite samples should be used as the
mean pH, instead of the calculated average of the
pH of small samples, and that the maximum and
minimum pH be used instead of calculated standard deviations.
(f) The foregoing applied especially to the
organic layer of the grey wooded soils.
REFERENCES
B ECKWITH, R. S. 1959 Titration curves of
soil organic matter. Nature 184: 745-746.
B OWSER, W. E., AND L EAT, J. N. 1958 Seasonal pH fluctuations in a grey wooded soil.
Can. J. Soil Sci. 38: 128-133.
pH OF GREY WOODED SOILS
R. F. 1950 Plants and Environment. J. Wiley & Sons, Inc., New York.
DOUGHTY, J. L. 1941 The advantages of a
soil paste for routine pH determinations.
Sci. Agr. 22: 135-138.
HANSEL, H. 1959 Die Verwendung der
Variationsbreite bei der Schiitzung der
Standardabweichung und bei der Varianzanalyse ungeordneter Blockanlagen, sowie
eine weitere Vereinfachung des Hartey's
Verfahren durch directe Bestimmung von
Grenzdifferenzen. Die Bodenkultur 10:
148-158.
MANNIGER, E. 1957 Untersuchungen fiber
DAUBENMIRE,
105
die bodenbiologischen bedingten periodischen Reaktions - Schwankungen der
Waldboden. Acta Agronomica (Budapest)
7: 217-228.
Moss, H. C., AND ST. ARNAUD, R. J. 1955
Grey wooded (podzolic) soils of Saskatchewan, Canada. J. Soil Sci. 6: 293-311.
SCHEFFER, F., AND ULRICH, B. 1960 Humus
und Humus diingung. Lehrbuch der Agrikulturchemie und Bodenkunde, III Teil, Band
I. S75.
C-J. AND CHINN, N. L. 1957 Direct
use of pH values in statistical analysis of
soil reactions. Soil Sci. 84: 219-224.