630.7
I£6b
no. 678
cop. 8
UNIVERSITY OF
AT nn
AT URBANA-CHAMPAIQN
AGRICULTURE
CIRCU UNIVERSITY OF JLLINOIS
-
CHEMICAK
as Aids in
BURNING HARDWOOD
TREE STUMPS
By C. S. WALTERS and K. R. PETERSON
Bulletin 678
UNIVERSITY OF ILLINOIS • AGRICULTURAL EXPERIMENT STATION
CONTENTS
Previous Studies and Reports 3
Combustion of wood 3
Fuelwood burned under controlled conditions 4
Burning proceeds by zones 4
Chemicals encourage combustion 5
History of chemical stump removers 6
Preliminary Tests of Chemicals 6
Chemical screening tests 6
Block tests 8
Field Tests — Methods and Results 10
Design of 1 956 stump tests 10
Design of 1 957 stump tests 13
Evaluation of burning tests 13
Amount of stump destroyed 14
Satisfactory-unsatisfactory assay of burning results 18
What is a satisfactory result? 19
Other Tests and Analyses 20
Relative importance of compounds in promoting combustion 20
Spectrochemical analyses of Formula 51, with and without surfactant. . . .21
Effect of surfactant on liquid absorption 22
Chemical dosage and stump size 23
Summary and Conclusions 27
Literature Cited . 28
ACKNOWLEDGMENTS
THE AUTHORS are indebted to the following persons for serving as evaluators in the
assay of stump-burning results: J. S. Ayars, J. D. Bilbruck, R. H. Brown, W. F. Bulkley,
R. J. Campana, J. C. Carter, F. M. Clark, L. B. Culver, T. W. Curtin, G. T. Frampton,
J. C. Gabbard, J. K. Guiher, E. B. Himelick, J. J. Jokela, V. L. Kretschmer, R. W.
Lorenz, R. O. Lyon, R. T. Milner, W. D. Murphy, D. H. Percival, R. G. Rennels, J. N.
Spaeth, and J. I. Zerbe.
Special acknowledgment is due to R. S. Chamberlin, Director of the Division of
Campus Development, Physical Plant Department, who not only served as evaluator,
but also arranged for the loan of drilling equipment and for fencing the test area.
We are grateful to H. W. Norton and R. D. Self for suggestions about the
statistical analyses; to J. S. Machin, Head of the Section of Physical Chemistry,
Illinois State Geological Survey, for advice about the spectrochemical analyses; and
to Juanita Witters, who made the analyses. T. W. Curtin and H. S. Scholten helped
bore the hundreds of holes required in the stump tests, a contribution which was
deeply appreciated.
Urbano, Illinois November, 1961
Publications in the Bulletin series report the results of investigations
made or sponsored by the Experiment Station
\o3(
INORGANIC CHEMICALS AS AIDS IN BURNING
HARDWOOD TREE STUMPS
C. S. WALTERS and K. R. PETERSON1
Ix THE PAST 10 YEARS phloem necrosis and Dutch elm disease have
killed thousands of elm trees in Illinois and other states east of the
Great Plains. Thousands more probably will die in the future, since
there is little hope that the two diseases can be controlled with the
methods currently in use. Many of the dead trees are in cities and
must be cut promptly to reduce the physical hazard, thereby leaving
an unsightly stump.
Tree stumps are difficult to extract from the ground because
Nature designs the cantilever system to resist the terrific forces that
are imposed at the groundline reaction point. Explosives and heavy
equipment cannot be used to remove the stumps on city property, and
digging them out is too costly and arduous. Burning stumps has been
tried, but for the most part the results have not been satisfactory for
the homeowner.
The burning process is not a direct one, but involves a series of
interdependent thermochemical factors. The conditions which favor
combustion of wood rarely are present in the environment of a stump,
and they are impractical for the homeowner to achieve.
There is thus a great need for a practical and economical way of
eradicating tree stumps on residential property. That is why we under-
took our investigation. Our principal objective was to test a limited
number of chemicals for their ability to promote glowing combustion
of stumpwood. We realix.ed that finding the most effective mixture
of chemicals was probably a matter of chance, since the environment
under which stumpwood is burned is highly variable.
Previous Studies and Reports
COMBUSTION OF WOOD
The literature concerning the burning of wood and how it is af-
fected by chemicals impregnated into the wood, dates back more than
400 years before the birth of Christ. At that time Herodotus reported
that the Egyptians steeped wood in alum solution to make it resistant
1 C. S. Walters, Professor, and K. R. Peterson, Assistant Professor, of Wood
Technology and Utilization.
4 BULLETIN No. 678 [November,
to fire (I).1 Since then, hundreds of reports have been written on the
subject. Most of them, however, have dealt with measures that tend
to make wood resistant to burning rather than with those that make
wood burn.
So far as we have been able to ascertain, the only data that have
been published on controlled stump-burning tests have been two re-
ports issued by the University of Illinois (10, 11). Other reports about
making wood burn deal almost exclusively with fuelwood, and par-
ticularly with seasoned wood.
FUELWOOD BURNED UNDER CONTROLLED CONDITIONS
According to Hawley (3), the combustion of fuelwood is controlled
by the size and distribution of the pieces of wood, by the air supply,
and by the way in which the heat of combustion is dissipated or re-
tained. When wood is burned in the fireplace, the air supply may be
regulated by the damper or by the size and arrangement of the sticks
placed in open piles. The dissipation of the heat is controlled by the
damper and the walls of the fireplace. This type of burning involves
an exothermic reaction in which the gases and vapors produced during
pyrolysis are burned, and charcoal is formed. The charcoal is then
burned to ash. When a large piece of wood is ignited, it burns as long
as the necessary heat penetrates from the outside to the inside of the
wood.
Fuelwood can be dried and burned under controlled conditions. It
is impractical to dry tree stumps in contact with soil, however, and the
conditions required for combustion of stumpwood are difficult for the
homeowner to control.2
BURNING PROCEEDS BY ZONES
Browne (1), in his very comprehensive report, discusses the differ-
ences between flaming combustion of wood and glowing combustion.
When wood is heated in air, the course of combustion is progressive,
proceeding through several stages which are determined by temper-
ature and which Browne refers to as zones. In Zone A, a stage with
temperatures below 200° C, pyrolysis is slow. The gases produced at
this stage of burning are not ignited.
Temperatures in Zone B range from about 200° to 280° C., a state
at which the mixture of gases produced still are not readily ignitible.
1 Numbers in parentheses refer to literature cited on page 28. The par-
ticular reference to Herodotus occurs on page 19 of the first citation.
1 A method for burning stumps in a metal "stove" or brick enclosure is de-
scribed in Mimeo F-261 (revised). A single copy is free from the Forestry
Department, 219 Mum ford Hall, Urbana, Illinois.
1961] BURNING HARDWOOD TREE STUMPS 5
The gases start burning, however, in Zone C (280° -—500° C), where
they evolve as a result of secondary pyrolysis and are ignited by a pilot
flame to burn outside the wood. The charcoal that has formed may or
may not burn, depending on whether a supply of oxygen is available
at the surface of the wood and whether enough heat penetrates the
layer of charcoal to advance the wood underneath to an exothermal
point.
In Zone D, at surface temperatures above 500° C., the layer of
charcoal on the outside of the wood burns, while the interior of the
wood still may be at various stages of burning, ranging from Zones
A to D. When the surface temperature rises over 1,000° C., the char-
coal (carbon) is consumed at the surface as fast as Zones A to C pene-
trate the wood.
CHEMICALS ENCOURAGE COMBUSTION
As already pointed out, charcoal needs both oxygen and heat
to glow. It is reasonable to assume that an adequate supply of oxygen
is available to the burning stump. Supplying enough heat is often
more of a problem. Theories of glow prevention indicate that the
presence of chemicals may improve conduction or absorption of heat
through the charcoal and wood. They may also lower ignition temper-
atures in the various zones and hasten the burning of carbon.
Browne points out that the net heat liberated by the combustion of
wood depends on the ratio of carbon monoxide to carbon dioxide in
the combustion products. Chemicals impregnated into wood, however,
could alter the ratio, thereby stimulating or retarding glowing com-
bustion. It appears that such chemicals could have a catalytic effect,
inhibiting the formation of carbon monoxide and promoting the
formation of carbon dioxide, or of carbon dioxide and hydrogen. In
such an event, combustion would be encouraged.
Although burning tree stumps may be a totally different problem
than destroying soot in heating systems, the two problems become
similar in nature when stumpwood is converted to charcoal, since both
soot and charcoal are forms of carbon.
Nicholls and Staples (9) concluded that the action of soot-removing
chemicals is restricted to metallic salts or those formed by burning
metals. The action of the chemicals was to lower the temperature at
which the soot burned.
The chlorides sublime at relatively low temperatures, and in the
Xicholls-Staples tests were particularly effective in lowering the igni-
tion temperature of soot. Some of the chlorides set free chlorine gas
6 BULLETIN No. 678 [November,
when burned, and the chlorine apparently increased the ease with
which the soot united with oxygen. Chlorides of manganese, iron, and
copper were particularly effective as soot burners. Lead salts, particu-
larly lead iodide, also gave good results in burning soot.
Nicholls and Staples concluded that oxygen had to be in the furnace
gases for a soot remover to be effective, and that the burning process
was limited to that of oxidation. They further concluded that the use
of salts high in oxygen, for example dioxides and chlorates, did not
increase the effectiveness of burning. In their studies the fuel ash
apparently was an effective aid in burning soot, since ashes high in
alkalies and metals were found to act as soot removers.
HISTORY OF CHEMICAL STUMP REMOVERS
One of the earliest known reports on the destruction of stumps
with chemicals was Coggins' account (2) of tests in which sulphuric
and nitric acids were used in varying proportions for destroying
stumps. Coggins concluded, however, that sound stumps could not be
destroyed with either or both acids, that the method was wasteful of
time, and that handling acid was dangerous.
In 1939 Mrs. A. Henn (4) recommended boring a vertical hole
1 inch in diameter in the center of a stump and placing 1 ounce of
potassium nitrate (saltpeter) in the hole. The hole was then to be
filled with water and sealed with a wood plug. The stump allegedly
burned, "roots and all," following ignition 6 or 8 months after
treatment.
Howell (5), Jackson (7), Morriss (8), and others have recom-
mended modifications of Henn's method. Both the original method
and some modifications were tested by the University of Illinois Agri-
cultural Experiment Station in 1949, and the results were reported
as unsuccessful (10).
Since 1953, approximately 400 water-soluble, inorganic compounds
or mixtures of them have been laboratory- or field-tested for their
ability to promote glowing combustion of wood. This report presents
the methods and results of the tests.
Preliminary Tests of Chemicals
CHEMICAL SCREENING TESTS
The compounds listed at the top of page 7 were chosen from hun-
dreds of chemicals that have been tested by the U. S. Forest Products
Laboratory (6) as fire retardants, because they showed the strongest
tendency to promote glowing combustion of wood.
1961} BURNING HARDWOOD TRKK Sir.Mi-s 7
Cupric chloride, CuCU Sodium molybdatt-, XajMoCX
Ferric chloride, FeCl, Lead acetate, Pb1OH(C3H3OJ)3
Manganese dichloridc, MnCU Cupric sulfate, CuSO,
Sodium dichromatc, NajCrjOi Chromium trioxide, CrO3
A 5-percent aqueous solution of each chemical was prepared. Sev-
eral drops of solution, or of a combination of solutions, were pipetted
on the center of a 9-centimeter disk of filter paper, and the paper was
dried at 105° C. for 1 hour. Each disk was placed on an asbestos-wire
gauze square and ignited. The flame was extinguished as soon as it
ignited the treated spot. Results of the burning tests were judged on
the ability of the chemical treatment to promote glowing combustion
that destroyed the treated portion of the paper disk. The results were
rated "good" (a vigorous reaction that destroyed the entire area),
"fair," or "poor." Each treatment was tested in triplicate and an
"average" rating was assigned the series of tests. Table 1 lists 23
formulations that received the "good" rating.
The nine formulas that appeared to give the best results were
retested in quintuplicate. In these tests the scoring ranged from 1
(fair) to 5 (excellent). The sum of the five numerical ratings was
Table 1 . — Chemical Composition of Formulas Rated "Good" in Paper-
Disk Assay of Their Ability to Promote Glowing Combustion
Percentage (by weight) of compound in formula
Formula PhOH
FeCl3 CuCl, Na2Cr207 /r „ A\ MnCl2 Na2MoO4 CuSO4 CrO3
C.
75 0 25 0
G. . . .
75 0
25 0
I
50 0 50 0
I
50 0
50 0
K
49 0
49 0 20
5
6 3
37 5
56 2
6
18 8
62 4
18 8
7
. . 33 0
34 0
33.0
8
62
37 6
56 2
13
. . 25 0
75 0
18
75 0
25 0
20
6 2
18 8
56 2
18 8
21
18 8
56 2
18 8
6 2
27... .
. . 56 2
6 2
37 6
28
25 0
75 0
31
25 0
75 0
36
56 2
31 3
12 5
43
25 0
75 0
45...
75 0
25 0
47
50 0
50 0
48
75 0
25 0
51
18 8
56 2
12 5
12 5
69..
37.6
56 2
6.2
BULLETIN No. 678 [November,
Table 2. — Rating Indexes and Chemical Composition of Nine
Formulations Rated Best in Paper-Disk Burning Tests
Formula
Rating
index"
Percentage
(by weight)
of compound in
formula
FeCl3
CuCl2
Na2Cr2O7
PbOH
(C2H302)3
MnCl2 Na2
MoO4
51
24
18.75
18.75
62 'SO
37.50
37.50
6.25
56.25
56
,25
12.50
56.25
31.25
12.
18.
12.
50
20
. . 23
6
56
18
6
6
56
18
25
25
75
25
,25
.25
.75
75
36
. . 21
is
56
^75
25
50
6
. . 20
5
. . 14
8
. . 14
56.25
37.50
18.75
27
10
21
9
6.
56.
25
69
9
37
.50
25 6.25
* Rating based on five burning tests scored as follows: Excellent, 5; very good, 4;
average, 3; good, 2; fair, 1.
used to rank each formulation. Table 2 shows the composition and
ratings for the nine formulations.
BLOCK TESTS
One hundred eight basswood (Tilia americana L.) blocks 1 inch
square and 4 inches along the grain were used to test two compounds
and two mixtures of compounds. Each chemical was tested in tripli-
cate, using three dosage levels and three diffusion periods.
The two mixtures and two compounds selected for testing were:
Formula 5 1,1 Formula K, sodium nitrate, and ammonium nitrate.
Formula 51 was selected because of its high rating in the screening
tests. Formula K and sodium nitrate were included because they had
produced good results in field tests, even though they had not appeared
in the top nine formulations. Ammonium nitrate was included because
it has a higher solubility rating than the sodium salt, and there was
reason to believe that it would yield satisfactory results.
A j^-inch hole 1 inch deep was bored parallel to the longitudinal
axis in one end of each block. The blocks were brought to "green"
condition by soaking 48 hours in distilled water, and the chemical was
placed in the hole.
The dosages were: 0.5 gram (0.0011 pound) per cubic inch of
block; 1.0 gram (0.0022 pound) per cubic inch; and none (controls).
The "green" blocks were placed in a pan containing about 1/2 inch of
water with the solid end in the water. Thirty-six blocks were removed
1 Subsequently manufactured as "Stumpfyre" under license from the Uni-
versity of Illinois Foundation, U. S. Patent No. 2,947,110.
7967]
BURNING HARDWOOD TREE STUMPS
o"2
!!;
x c>o — x~>
IA. 5 o> x t» oo
1XI/XO "1 CN O »f <*!") f CS*X
£
o >«
_X>OCu-. X'tX 1/5 m «/5 O O "5 <S l^CT-'C^-CM \C >r> •*"*• >r> C O* O
5 — — »« (s!^« t<5« « — CJ »•
i
U
E.
E -§=
E" " E
(£ i
s
C ^•l^iA^fO^COC OCO***^^Or^ *f<*/;oO"t'XCO •t't^^'f^'i/^^^'t'O fO
r^^"^f*5f O^OC*
•ta OOO^^T^O'O *OXi/*t^C • *^ ^'OCp*5ClO'O'O O'Of^'^^^O Cf^fSOOoC^^'^-(
fi §f
^1
"S Ps*t-">r^iOO**1C t*» !•* ^ t** »* i/; **) P*I C4 i/; *f f-» X "5 f^io^^fr^i^O1 *'O>OP^oC^*f*iOfs*O "^
K^ — CN«(S-*(N (NfN — ~f5<N<N ^t (N — (N — — (S ~ ^ ~ f*J <N — ^ f-)^ — — (N — CS — <N (N
||
|l
1 1 |i
::::::E ;;:;:: a :;;;:: E • c 25
Ed
I Is5 -*22SS w>-'
10 BULLETIN No. 678 [November.
at the end of 2 days, air-dried, and burned. Other groups were burned
after 5-day and 8-day diffusion periods.
Each air-dried block was ignited with a Bunsen-burner flame ap-
plied to the solid end for 5 minutes. The results were rated "good,"
"fair," or "poor," depending upon how much of the block was burned.
The results of the paper-disk and wood-block tests were used to
design a field test.
Field Tests — Methods and Results
DESIGN OF 1956 STUMP TESTS
Twenty-four American elm (Ulmus americana L.) stumps, ranging
from 12 to 42 inches in diameter, were treated and burned in a field
test of the two mixtures and two compounds described in the pre-
vious section. In addition, eight untreated stumps were included in
the burning tests (Table 3).
All stumps were cut as close to the ground as practical. The
"average" diameter,1 perimeter, and cross-sectional area of each
stump were measured. Perimeter measurements followed the outline
of the flat surface of the root extensions. It was assumed that each
stump was 12 inches deep, including the inch or two aboveground.
Although some stumps may have extended deeper than this, destruction
of the top 12 inches would usually constitute satisfactory results.
Vertical holes 2 inches in diameter and about 6 inches deep were
bored with a twist drill in all 32 stumps (Fig. 1). The holes were
spaced 4 inches apart, using a hardboard template to locate the centers.
The holes were freed of chips before the chemicals were placed in
them.
Treatments were randomly assigned to stumps. Dosage originally
was based on stump diameter, but this basis was abandoned because
of the inaccuracy involved in determining "average" diameter for
stumps with irregular, cross-sectional shapes. The dosage used was
about 1 gram (0.0021 pound) of chemical per cubic inch of wood
contained in the top 12 inches of the stump.
Chemical mixtures were blended in a large feed-mixing machine.
The total chemical for each stump to be treated was weighed and
distributed more or less equally among the holes about the third week
of June. The holes were then filled with water. All stumps, treated
and untreated, were covered with reflector shields (Fig. 2) to protect
them from rain during the diffusion period.
1 The average of two diameters, one measured at right angles to the other.
1961}
H. \KD\VI » >n TKKK S
11
Most of the aboveground portion of this stump was removed with a power
saw. Two-inch holes for the chemical were bored to a depth of about 6
inches. Some of the holes were not close enough to the bark, and an extra
hole or two in some areas would have been desirable. (Fig. 1)
This reflector shield is about 4 feet square, made of hardboard and framed
with pine strips; the under (reflecting) side is covered with household-type
aluminum foil fastened with pressure-sensitive paper tape. The supporting
stakes should have been heavier. The shields were kept as close to the stump
as practical, as shown by those in the background. (Fig. 2)
12
BULLETIN No. 678
[November,
Dry kindling was piled on each stump and ignited. A liberal amount was
used, varying with the cross-sectional area of the stump. (Fig. 3)
The experiment had been designed to test 2-week and 4-week
diffusion periods. Rainy weather, however, made burning impractical
at the end of 2 weeks, so this part of the experiment was abandoned.
After 4 weeks the weather was again rainy. In addition, the holes in
the treated stumps were full of liquid, while those in the untreated
stumps contained little if any. The liquid resulted from the hydro-
scopic nature of the chemical, not rain. Attempts to remove it with
wicks of paper toweling were unsuccessful. The combination of the
rainy weather and the liquid in the treated stumps made it imprac-
tical to burn the stumps after the 4-week diffusion period, so this
part of the experiment was also abandoned. Finally, most of the liquid
was removed with a battery syringe about 3 months after the stumps
were treated.
A day or so after the liquid was removed, the stumps were burned.
First, the reflector shields were removed, then a liberal amount of
kindling was piled on each stump and ignited (Fig. 3). About 2 hours
after the kindling was ignited, it had burned to a bed of coals, and the
reflector shields were replaced over the stumps.
1961] BURNING HARDWOOD TRKK STUMPS
Table 4. — Size and Treatment of American Elm Stumps With
Formula 51, and Evaluation of J957 Burning Tests
13
Stump
No.
Surface
area
No. of
holes
Amount
of
chemical
Volume
Reflector destroyed
shield by
burning"
Satis-
factory-
ratings
awarded
1..
Sq. in.
511
16
Pounds
4 00
Percent
Yes 85
Percent
94
8
. 593
21
5.25
93
100
12
. 237
9
2.25
78
81
14
. 397
12
3.00
88
94
19
695
14
3.25
66
62
Average
. 487
14
83
86
5 .
522
15
3.75
No 94
100
7
. 314
11
2.75
42
37
10
. 644
16
4.00
93
100
13
554
13
3.25
94
100
20
. 672
25
6.25
47
6
Average. . . .
2
. 541
536
16
16
None
78
Yes 16
69
0
3
415
14
37
25
11
303
10
"
11
6
16..
. 741
21
«
89
100
17
442
12
"
29
0
Average. . . .
4..
. 487
503
15
18
None
36
No 25
26
0
6
. 467
16
19
0
9
. 348
12
• •
30
0
15
. 370
13
"
31
6
18
. 1,036
29
n
36
12
Average. . . .
Average all
stumps. . .
. 545
. 515
18
16
28
4
• Average for each estimated percentage of original volume. Average for each stump is
for 16 estimates. Percentages obtained by decoding angular transformation of data.
DESIGN OF 1957 STUMP TESTS
Treatment in 1957 was about the same as in 1956, with the follow-
ing exceptions: Only Formula 51 was used, and the test was designed
to show the effect of the chemical alone, the shields alone, and the
interacting effect of the two together (Table 4). The dosage was also
smaller, being about 115 grams (0.25 pound) of chemical per hole, or
0.27 gram (.0006 pound) per cubic inch of wood in the top 12 inches
of the stump. Twenty American elm stumps, including five control
stumps, were in the tests.
EVALUATION OF BURNING TESTS
At the end of the burning period, about 2 weeks, the ash was
removed from the stump remains and evaluators were asked to judge
the results by estimating the percentage of each stump destroyed. Each
14 BULLETIN No. 678 [November,
evaluator also indicated whether the results were "satisfactory" or
"unsatisfactory," basing his decision on the assumption that the stump
was located on his home grounds.
The term "stump" was identified for the evaluators as ". . . includ-
ing the wood left aboveground, the projection of the cross-section to
a depth of not more than 12 inches below ground, and the side roots
to a depth of 4 inches."
In 1956 half of the 20 evaluators were "foresters" and half were
"homeowners." Foresters were members of the Forestry Department.
Homeowners were members of other departments of the University or
of the Botany Department, Illinois State Natural History Survey. The
classification was made because most of the foresters were generally
familiar with the tests, had actually removed stumps from their home
grounds, and were believed to be more familiar with the root structure
of trees. The authors did not participate in the evaluations.
In 1957, 16 evaluators judged the results of burning tests. The
evaluators were not classified as they were in 1956.
The percentages of wood destroyed, as estimated by the evaluators,
and their ratings of "satisfactory" or "unsatisfactory" were used as
indexes of the effectiveness of treatment. Angular transformations
were made for the percentage data before they were analyzed by an
analysis of variance. The "satisfactory-unsatisfactory" ratings were
coded "1" and "0," respectively, and the coded data were analyzed by
an analysis of variance.
AMOUNT OF STUMP DESTROYED
1956 tests. Table 5 shows the average volume of stump burned in
1956, expressed as a percentage of the original volume. Table 6 is an
analysis of variance for angular transformations of the percentage
data.
In percentage of wood destroyed, Formula 51 was the best treat-
ment. Average percentage for Formula 51 was 83; for Formula K,
57; for sodium nitrate, 53; for ammonium nitrate, 41; and for un-
treated stumps, 37. Standard errors for the averages were less than
1 percentage point. The "treatments" component was significant at
the 5-percent level of confidence (Table 6).
Homeowners' estimates of the volume destroyed averaged 55 per-
cent for all treatments; foresters' estimates, 53 percent. The difference
between these means was not significant, as shown by the nonsignifi-
cant variance ratio for "Groups" in Table 6. This is why evaluators
were not grouped when the 1957 tests were designed.
The percentages for stumps within a treatment varied significantly.
1961] BURNING HARDWOOD TRKE STIMI-S 15
Table 5. — Average Volume of Elm Stumps Burned, as an Estimated
Percentage of Original Volume, 1 956 Tests
Percentage as estimated by —
Homeowners
Foresters
Formula 51
82
84
83
Formula K
57
58
57
Sodium nitrate
55
50
53
Ammonium nitrate
44
39
41
None (controls)
38
35
37
Average
55
53
* Percentages obtained by decoding angular transformations. Final averages for chemicals
ba>ed upon 120 observations; final average for controls based upon 160 observations.
Table 6. — Analysis of Variance for Angular Transformations of Data
Concerning Percentage of Stump Destroyed in 7956 Burning Tests
Source of
variation
Degrees of Mean
freedom square
Variance
ratio (F)
Signifi-
cance*
Treatments (T)
4
15,717
217
139
4,001
1,334
2,208
460
91
85
62
3
3
1
64
14
24
5
1
1
9
5
5
5
7
3
1
5
4
*
NS
NS
*
*
*
*
NS
Groups (G)
1
TG
4
Stumps, in treatments (S)
27
Individuals, in groups (I)
18
Homeowners
(9)
Foresters
(9)
TI
72
GS
27
SI (error)
486
Total
639
• *, significant at .05 level; **, significant at .01 level; ***, significant at .001 level;
NS, not significant.
Figure 4 shows the differences in opinions expressed by the various
evaluators when they examined Stump No. 1. Evaluator No. 7 esti-
mated that only 23 percent of the stump's original volume was de-
stroyed, whereas Evaluator No. 17 judged that the entire stump was
destroyed. Evaluators 2, 3, 4, and 14 all estimated that 80 percent of
the stump was destroyed, but they disagreed as to whether 80 percent
destruction was an acceptable result. Although Evaluator No. 2
thought the results were satisfactory, the other three rated them
unsatisfactory.
In the analysis of variance, individual evaluators within a group
("I" in Table 6) differed significantly in their opinions as to the
amount of wood destroyed. The significance was at the 0.001 level of
reliability. When the whole group was divided into "Foresters" and
"Homeowners," individuals within each subgroup differed significantly
16
BULLKTIN No. 678
[Norember,
1234567 89 10
HOMEOWNERS
II 12 13 14 15 16 17 18 19 20
FORESTERS
How 20 evaluators judged Stump No. 1 as to percentage destroyed by burn-
ing and as to whether the result was satisfactory or unsatisfactory. (Fig. 4)
at the same level. This means that the homeowners disagreed among
themselves as to the amount of wood in each stump that had been
destroyed by burning, and so did the foresters. However, the test of
significance indicated that variation was greater among the home-
owners. These results could, of course, have been reasonably pre-
dicted. One would expect individuals within a group to vary in their
evaluations. One would also expect the variation to be less among the
foresters, since they are generally more experienced in estimating the
cubic volume of geometric forms and they are more familiar with
the underground form of tree stumps.
The only interaction of factors that was statistically significant was
the interaction of treatment and individuals ("TI," Table 6). That is,
the estimates of individual evaluators varied with a change in treat-
ment. In other words, the evaluators were not consistent in their
evaluations of the different treatments. No bias was involved, for the
evaluators had not been told how each stump was treated. The incon-
sistencies might be expected, however, from the wide range in the
1961] BURNING HARDWOOD TREE STUMPS 17
burning results. It was much easier for an evaluator to estimate the
volume of wood destroyed as the percentage approached 100, complete
destruction, than it was to estimate the percentage when only a small
portion of the stump had been burned. Since the evaluators had not
seen each stump before it was burned, individual concepts of the
original form undoubtedly contributed to the variation among estimates.
1957 tests. Table 7 is a summary of the results of the 1957 burn-
ing tests. The average percentage of volume destroyed by burning
was 80 for elm stumps treated with Formula 51 and only 32 for stumps
containing no chemical. An analysis of variance (Table 8) indicated
that the effect of the chemical was highly significant.
Reflecting shields were used because they were believed to stimulate
burning. The variance ratio in Table 8, however, shows that the effect
of shields was not significant. Although the reflecting properties of the
shields apparently did not aid burning, the shields did protect the
smouldering fire from rain, and they made the burning stump less
hazardous to children and animals that frequented the test area.
The test of the CS, chemical by shield, interaction also proved
nonsignificant (Table 8). Thus, it appears that the chemical made the
real contribution to the burning process.
Table 7 . — Average Volume of Elm Stump Destroyed
as Percent of Original Volume, 7 957 Tests
Percentage destroyed
i rear, mem
Shield
No shield
./Average
Chemical (Formula 51)
83»
78»
80
No chemical
36"
28*
32
Average
61
54
57
• Based on 80 observations, or examinations of 5 stumps per treatment by 16 persons.
Percentages obtained by decoding angular transformations.
Table 8. — Analysis of Variance for Angular Transformations of Data
Concerning Percentage of Stump Destroyed in 7957 Burning Tests
(Treated Stumps Contained Formula 51)
Source of
variation
Degrees of
freedom
Mean
square
Variance
ratio (F)
Signifi-
cance*
Chemical (C)
1
1,111,090
21.14
***
Shield (S)
1
20,995
0.40
NS
CS
1
541
0.01
NS
Error
16
52,555
Total
19
significant at the .001 level; NS, not significant.
18 BULLETIN No. 678 [November,
SATISFACTORY-UNSATISFACTORY ASSAY OF BURNING RESULTS
1956 tests. Table 9 shows the percentage of total ratings judged
satisfactory in the assay of the 1956 burning results. Table 10, an
analysis of variance for the data, shows "Treatment" was a very highly
significant source of variation. Seventy-one percent of the ratings
awarded to the results obtained with Formula 51 were classified satis-
factory (Table 9). The next best rating was given the sodium nitrate
treatment, 31 percent. Only 15 percent of the ratings for stumps
burned without a chemical were classified as satisfactory.
Foresters called 30 percent of the ratings satisfactory; homeowners,
28 percent (Table 9). This difference was not significant (Table 10).
The GS, group by stump, interaction was very highly significant.
This means that the groups were not consistent in awarding ratings to
the different stumps within a treatment. In addition the TI, treatment
Table 9. — Percentage of Total Ratings Judged Satisfactory,
in Assay of 1956 Stump-Burning Tests
Percentage judged satisfactory by —
Homeowners
Foresters
Formula 51
67
75
71
Formula K
22
25
23
Sodium nitrate
30
32
31
Ammonium nitrate
13
5
9
None (controls)
12
17
15
Average
28
30
• Percentage for each chemical based upon 120 observations; average for controls based
ujion 160 observations.
Table 10. — Analysis of Variance for Coded0 Ratings
Awarded in Satisfactory-Unsatisfactory Assay
of Stump-Burning Results, 1 956 Data
Source of variation
Degrees of
freedom
Mean
square
Variance Signifi-
ratio canceb
Treatment (T) . ...
4
7.32
0.08
0.12
1.78
0.42
0.36
0.48
0.92
1.14
0.02
430 . 9 *•
4.5 N
7.1 N
104.8
24.8
21.4
28.3
54.2
67.2
t*
S
S
Groups (G)
1
TG
4
Stumps, in treatments (S). . . .
Individuals, in groups (I). . . .
.... 27
18
Homeowners
(9)
Foresters
(9)
TI
72
GS
27
SI (error)
486
Total
639
• Satisfactory ratings coded "1"; unsatisfactory ratings coded "0."
b ***, significant at .001 level; NS, not significant.
1961} BURNING HARDWOOD TRKK STIMI-S 19
by individual, interaction, was significant at the 0.001 level of prob-
ability. This means that the individual evaluators were not consistent
in assaying results as they judged the various treatfhents.
1957 tests. Table 11 shows the average ratings awarded the two
variables tested in the 1957 tests — chemical and shield. Table 12 is
an analysis of variance for the data. Essentially the same results are
shown by the ratings as were shown in Table 7 by the percentage data,
and the same conclusions are drawn from the data and their analysis.
WHAT IS A SATISFACTORY RESULT?
The results discussed in previous paragraphs raise the question as
to whether there was a minimum amount of wood that had to be
destroyed for results to be judged "satisfactory." In other words, was
there a correlation between the percentage of stumps destroyed and tile-
awarding of a satisfactory rating?
Figure 5 shows the relationship between the number of satisfactory
and unsatisfactory ratings awarded 1956 and 1957 tests and the
volume of stump destroyed. Frequency distributions for both kinds of
ratings are shown with their straight-line equations. As one would
expect, the frequencies for the unsatisfactory ratings were highest
when the results were poor and only a small portion of the stump was
Table 11. — Percentage of Total Ratings Judged Satisfactory
in Assay of 1957 Stump-Burning Tests
Percentage judged satisfactory
Shield
No shield
ttverage
Chemical
86»
69"
78
No chemical
26a
4"
19
Average
59
46
Percentage based on 80 ratings; 16 individual observations of each of 5 stumps.
Table 12. — Analysis of Variance for Coded" Ratings
Awarded in Satisfactory-Unsatisfactory Assay
of Stump-Burning Results, 1957 Data
Source of
variation
Degrees of
freedom
Mean
square
Variance
ratio (F)
Signifi-
cance1'
Chemical .
1
500 0
20 1
*•*
Shield
1
51.2
2.1
NS
CxS
1
12 8
05
\S
Error
16
24.9
Total
19
• Satisfactory ratings coded "1"; unsatisfactory ratings coded "0."
b ***, significant at .001 level of probability; NS, not significant.
20 BULLETIN No. 678 [November,
90
4 UNSATISFACTORY RATING
80 h
|70
*60
&
>. 50
| 40
S
£30
20
10
o SATISFACTORY RATING
f a"~1
0 10 20 30 40 50 60 70 80 90 100
VOLUME OF STUMP DESTROYED, PERCENT OF ORIGINAL VOLUME
Number of satisfactory and unsatisfactory ratings awarded as a function of
the volume of stump destroyed, 1956 and 1957 data combined. (Fig. 5)
destroyed. The reverse was true for the satisfactory ratings, with
frequencies being highest when results were good and large portions
of the stumps were destroyed.
The linear regressions for the two sets of data cross at about 62
percent. Thus, it appears that at least two-thirds of the stump had to
be destroyed before most of the e valuators were willing to rate the
results as satisfactory.
Individuals, of course, may disagree with the "average" results, as
they did in the study. For example, four ratings plotted in Figure 5
classified stumps only half destroyed as "satisfactory," and 10 ratings
classified stumps that were 85 percent destroyed as "unsatisfactory."
Other Tests and Analyses
RELATIVE IMPORTANCE OF COMPOUNDS IN PROMOTING GLOWING COMBUSTION
Twenty-three formulations that were rated "good" in the paper-
disk assays of their ability to cause glowing combustion (Table 1) were
examined to see whether one compound was superior to another in this
property.
Eight compounds were used in the 23 mixtures (Table 13). Sodium
dichromate not only appeared in 11 mixtures, but also was used in
greatest quantity. Next, on the basis of quantity used, was lead acetate,
which appeared in 10 mixtures. Cupric chloride and ferric chloride
were also used in 10 mixtures, but in smaller quantities. Formula 51
contained four of the first five compounds listed in Table 13.
1961] BURNING HARDWOOD TREE STUMPS 21
Table 13. — Relative Importance of Compounds in 23 Formulations0
Rated "Good" in Paper-Disk Assay of Ability
to Cause Glowing Combustion
Compound
Frequency of
appearance in
formulations
Total parts,
by
weight
Percent
of total
parts
Percent of
Formula 51,
by weight
Na.Cr.O7..
11
551.8
24.0
56 2
PbOH(CsH3Oo)s
10
437.5
19 0
12 5
CuClj
10
362.5
15 8
18 8
FeCU
10
301 7
13 1
MnCl.
8
256 2
11 1
12 5
CuSO4
4
199 0
8 7
CrO,
4
127.0
5.5
NaMoO4
3
64 2
2 8
Total
2,300.0
100.0
» See Table 1 for components in formulations.
SPECTROCHEMICAL ANALYSES OF FORMULA 51, WITH AND WITHOUT SURFACTANT
A number of short-term laboratory experiments were made to learn
more about the components in Formula 51 that diffused into the wood.
Two 6-gram samples of "Stump fy re" (a commercial product with
the same composition as Formula 51 )x were dried to constant weight
at 105° C. and reweighed. The dried samples were diluted with dis-
tilled water and filtered. The resulting precipitates and filtrates were
dried to constant weight and weighed again. Weights of the filtrates
and precipitates were expressed as percentages of the oven-dried
weights of the samples. Spectrographic analyses were then made of
the filtrates and precipitates.
A third 6-gram sample of the commercial mixture was dried to
constant weight, then divided into two subsamples of about equal
weight. The subsamples, designated "A" and "B," were diluted with
hot distilled water. To Sample B we added 1 milliliter (0.9 gram) of
a commercial surfactant,2 to determine whether it would improve the
liquid's "wettability" rating and its flow into the wood. The two sub-
samples were filtered and the filtrates and precipitates were dried to
constant weight for spectrographic analysis.
Results from only the third 6-gram sample are presented in this
report (Table 14). The spectrographic analyses for all three sets of
filtrates and precipitates were fairly consistent for lead, sodium, and
manganese. Some differences occurred, however, in the analyses for
chromium and copper. These differences were attributed to sampling
1 See footnote 1, page 8.
'This surface active agent was a liquid household detergent, containing a
mixture of the ammonium salt of a modified alkyl sulfate and an alkyl cthanol
amide.
22 BULLETIN No. 678
Table 14. — Percenf of Metal Present in Soluble and Insoluble Portions
of Formula 51 , With and Without a Surfactant
Percent in filtrate Percent in precipitate
Metal
Surfactant No surfactant Surfactant No surfactant
Cr .
3-30(X )
3-30(X-)
3-30(X + )
3-30(X + )
Cu .
3-30(2X)
3-30(2X-)
Mn
1-10(X )
I-10(X )
3-30(6X)
3-30(5X)
Na ....
3-30(2X)
3-30(2X)
Pb
T*
T*
1-10
1-10
• T = less than 1 percent.
error, since the commercial preparation from which the samples were
drawn probably was not a homogeneous one.
Shown in Table 14 are the ranges in metal content of the nitrates
and precipitates from the subsample without a surfactant and from the
one with a surfactant. The two sets of nitrate values are about the
same, as are the values for the precipitates. Adding the surfactant
apparently had little effect on the movement of metal ions.
A rough quantitative analysis is also shown for each cell in Table
14. For example, the percent of manganese in the insoluble precipitates
ranged between 3 and 30 percent, with the average amounts being
about 5 or 6 times (5X and 6X) the average amounts in the nitrates.
The insoluble precipitate amounted to about 20 percent of the
weight of an undried sample, and the nitrate about 60 percent. The
remaining 20 percent probably was mostly water, although chlorine
in the form of hydrogen chloride, as well as other components, may
have evaporated during the drying process.
EFFECT OF SURFACTANT ON LIQUID ABSORPTION
As described above, one sample of Formula 51 was divided into
two subsamples which were diluted in hot water; and a commercial
surfactant was added to one of the subsamples. These liquids are
referred to as Liquid A (without the surfactant) and Liquid B ( with
the surfactant).
Ten wafers of basswood about 2 inches square and J4 inch along
the grain were dried to constant weight at 105° C. Half were soaked
for 30 seconds in Liquid A and half, in Liquid B. Each wafer was
weighed before and after soaking and again after it was returned to
oven-dry condition. Average absorptions were determined for each
group of wafers.
The surfactant increased the rate at which the liquid penetrated
the wood. Average absorption of Liquid B was 1.545 grams, or more
1961] BURNING HARDWOOD TREK STUMPS 23
Table 15. — Effect of Surface Active Agent on Absorption
of Liquid Form of Formula 51 by Wafers of Basswood
Liquid"
Average weight of wafers (gm.)b
Average absorption (gm.)b
Untreated,
oven-dry
Freshly
treated
Treated,
oven-dry
Liquid
l)r>
chemical
A. .
5.552
6.315
7.230
5 618
5 . 805
0.763(x)
1. 545(2. 02x)
0.066(y)
0. 120(1. 82y)
B
5 . 685
• Liquid A: 10 gm. commercial grade of Formula 51 and 100 ml. water. Liquid B:
same as A plus 1 cc. surfactant (see footnote 1, page 21).
b Based on measurements for six specimens per treatment.
than twice that of Liquid A (Table 15). When absorption was cal-
culated on the basis of dry chemical, retention was 0.120 gram — again
about double the value for Liquid A.
CHEMICAL DOSAGE AND STUMP SIZE
Dosage in laboratory tests on small blocks ranged from 0.5 to 1.0
gram (0.0011 pound to 0.0022 pound) per cubic inch of wood. Average
dosage for the 1956 stump tests was 1 gram (0.0022 pound) per cubic
inch of wood in the top 12-inch portion of the stump; and for the
1957 treatment, 0.27 gram (0.0006 pound) per cubic inch. The results
of burning indicated that these amounts were adequate for Formula
51. The question arose, however, as to which of several measurements
should be used as the independent variable in correlating stump size
with chemical dosage.
Diameter, perimeter, top area, and number of holes bored in the
stump could be used as indexes of stump size. Of these, top area
appeared to be the most satisfactory index, although it was somewhat
less practical to measure than "average diameter," which required two
measurements at right angles to each other. The measurement of
diameter was not always precise when stumps were irregular in cross
section (Fig. 6), and we did not knowr whether there was good correla-
tion between average diameter and the cross-sectional area of the top
for all stumps. Dosage was therefore based on top area.
Figure 7 shows the relationship between top area and average
diameter (Table 3). Figure 8 shows stump perimeter in relation to
average diameter. The correlation between surface area and average
diameter (r = 0.917) was much better than that between perimeter and
average diameter (r = 0.761).
The equation derived for calculating the surface area in square
inches is Y = 39.3X — 273, with Y equaling the area and X, the
24
BULLETIN No. 678
[November,
The irregular pattern of many American elm stumps makes it difficult to
correlate size and chemical dosage. The stump which occupied this spot was
about 42 inches across at its widest point. (Fig- 6)
1400
1300
1200
1100
1000
900
s800
< 700
E
t 600
P
500
400
300
200
100
Y-39.3X-273
r • 0.917
I I I I I I
10
15 20 25 30
AVERAGE DIAMETER, INCHES
35
40
Relationship between cross-sectional area of stump top and the average of
two diameter measurements made at right angles to each other. (Fig. 7)
1961]
BURNING HARDWOOD TRKE STUMPS
25
average diameter. Thus, the surface area of a stump having an average
diameter of 20 inches would be expressed as Y == (39.3) (20) — 273,
or 513 square inches. The difference between this value and the area
of a 20-inch circle (314 inches) is due to the irregular pattern of the
stump's top caused by the root swell.
Figure 9 shows the relationship between the number of holes bored
in the 1956 and 1957 test-stumps and cross-sectional area of the stump
top. As shown in Table 3, the number of holes ranged from seven for
a stump 12 inches in diameter (156 inches of top area) to 59 for a
42-inch stump (1,248 square inches of top area). We believe that the
minimum number of holes needed for effective treatment is deter-
mined by the equation, Y = .042X — 1.1, in which Y equals the
number of holes and X, surface area in square inches.
Figures 7, 9, and 10 will enable one to determine dosage for round
or irregularly shaped stumps. If a stump is fairly round, average
diameter can be determined, this can be converted to cross-sectional
25
20
K 15
r * 0.761
10
15 20 25 30
AVERAGE DIAMETER, INCHES
35
40
Relationship between stump perimeter and average diameter of stump
top. (Fig. 8)
26
BULLETIN No. 678
[November,
59
50
s45
§ 40
§-
t 30
a. 25
jjao
15
10
5
Y = .042X — 1.1
r = .922
I I I I I I I I I I I I I I I
100 200 300 400
500 600 700 800
TOP AREA, SO. IN.
900 1000 1100 1200 BOO
Relationship between cross-sectional area of stump top and number of holes
bored, 1956 and 1957 data combined. (Fig. 9)
20
10
I I I I
Y = 0.6 + .0102X
r = .479
I I I I I I I I I I
200
400
600 BOO 1000
TOP AREA SO. IN.
1200
1400
1600
Amount of Formula 51 applied to stumps as a function of size (cross-
sectional area of stump top). (Fig. 10)
area of the top by use of Figure 7, and correct dosage can then be
determined from Table 10. If the stump is irregularly shaped, holes
can be bored on 4-inch centers, the surface area can be determined by
using Figure 9, and the amount of chemical needed, by using Figure 10.
1961] BURNING HARDWOOD TREE STI MI-S 27
Summary and Conclusions
Stumps that remain from cutting trees are difficult to remove from
residential areas because explosives or heavy machinery cannot he-
used, and hand methods are too arduous and expensive. There is a
great need in Illinois and elsewhere for a practical and economical
means of removing the stumps of disease-killed elm (Ulmiis spp.)
trees. Commonly recommended methods involving the use of potas-
sium nitrate (saltpeter) to promote combustion have not achieved
satisfactory results. Other compounds known to cause glowing com-
bustion of wood in fire-retardant studies were screened for use in
burning tree stumps.
Eight compounds, as well as over 400 combinations of them, wen-
tested for their ability to promote glowing combustion of paper disks.
The most promising combinations were further tested with small
blocks or wafers of wood. Finally, in 1956, 24 American elm (Ulmits
amcricana L.) stumps were treated either with ammonium nitrate,
sodium nitrate, Formula K, or Formula 5 1.1 The stumps were ignited
about 3 months after treatment. Eight untreated control stumps were
included in the burning tests.
Results of the burning tests were judged by 20 evaluators, who
estimated the percentage of wood destroyed and rated each test a>
"satisfactory" or "unsatisfactory." Formula 51, a mixture of 4.5 parts
sodium dichromate, 1.5 parts cupric chloride, 1 part lead acetate, and
1 part manganese dichloride, gave superior results, destroying, on the
average, 83 percent of the stump. The average percentage value for
Formula K was 57; for sodium nitrate, 53; for ammonium nitrate, 41;
and for untreated stumps, 37.
In 1957. a second series of experiments with 20 American elm
stumps tested the interacting effect of Formula 51 and reflecting shields
on burning. The shields had an insignificant effect on burning results,
whereas the chemical had a highly significant effect. The interacting
effect of using both chemical and shields was not significant. In these
tests the average percentage of stump destroyed was 80 for stumps
treated with Formula 51, and 32 for the untreated ones.
Of 120 ratings given to stumps burned after treatment with For-
mula 51 in 1956, 71 percent were classified as satisfactory. The next
best treatment — sodium nitrate — was considered satisfactory in only
31 percent of the ratings. In 1957, 78 percent of 80 ratings for treated
stumps were satisfactory, whereas only 19 percent of the ratings for
stumps burned without chemical treatment were satisfactory.
'Subsequently named Stumpfyre, U. S. Patent No. 2,947.110.
28 BULLETIN No. 678
The relationship between number of satisfactory ratings and
amount of stump destruction showed that at least 62 percent of a
stump had to be destroyed before most evaluators were willing to rate
results as satisfactory.
Twenty-three of the best chemical formulations were examined to
see whether one compound was more important than another in pro-
moting combustion. Sodium dichromate appeared most frequently,
followed by lead acetate, cupric chloride, ferric chloride, and manga-
nese dichloride. Formula 51 contained four of these compounds.
When a surfactant was added to an aqueous dilution of Formula
51, basswood wafers absorbed 2.02 times as much liquid and 1.82
times as much dry chemical as wafers treated in wafers containing no
surfactant. The surfactant also increased the rate at which the liquid
penetrated the wood.
First-order equations were calculated for several relationships con-
cerning stump size (in terms of cross-sectional area or number of holes
bored on 4-inch centers) and chemical dosage.
Literature Cited
1. BROWNE, F. L. Theories of the combustion of wood and its control.
U. S. Forest Prod. Lab. Rpt. 2136. 1958.
2. COGGINS, H. C. Destroying stumps with acid. Agr. Gaz. N. S. Wales
24 (11):967. 1913."
3. HAWLEY, L. F. Combustion of wood. Wood Chemistry, L. E. Wise,
editor, pp. 671-679. Reinhold Pub. Corp., New York, N. Y. 1944.
4. HENN, A. Stump remover. Prairie Farmer 111 (26):11. 1939.
5. HOWELL, H. A., Arkansas extension forester, in a letter to senior
author dated September 23, 1947.
6. HUNT, G. M., TRUAX, T. R., and HARRISON, C. A. Experiments in fire-
proofing wood, third progress report. Amer. Wood Preservers'
Assoc. Proc. 28:71-93. 1932.
7. JACKSON, W. E., Kentucky extension forester, in a letter to senior
author dated July 29, 1947.
8. MORRISS, R. H. Destroying stumps with saltpeter. U. S. Dept. Agr.
Forest Serv. Random Forestry Note, p. 9. May 27, 1947.
9. NICHOLLS, P., and STAPLES, C. W. Removal of soot from furnaces and
flues by the use of salts or compounds. U. S. Dept. Commerce Bur.
of Mines Bui. 369. 1932.
10. WALTERS, C. S. The value of chemicals for eradicating stumps by
burning. Arborists' News 14 (4): 30-34. 1949.
11. - — , Fox, H. W., and KULP, D. A. A progress report on the
use of chemicals in destroying hardwood tree stumps by burning.
111. Agr. Exp. Sta. Note No. 2. 1948.
3500—11-61—74983
UNIVERSITY OF ILLINOIS URBANA
Q.630.7IL6B COOS
BULLETIN URBANA
6781961
30112019530424