General-buffer catalysis of the reaction of N-(hydroxymethyl)benzamide: A new pathway for the aqueous reaction of carbinolamides

Article abstract of DOI:10.1016/j.tetlet.2005.03.011

The second-order rate constants for the general-buffer catalyzed breakdown of N-(hydroxymethyl)benzamide (1) in water at 25°C, I = 1.0 (KCl) by pivalic, acetic, chloroacetic, and dichloroacetic acid were determined by initial rates. The observed rate increased with increasing amounts of the acidic form of the buffer and a Br?nsted correlation of α = 0.35 was determined. The results presented here, represent the first evidence for a general-buffer catalytic mechanism for the aqueous reaction of 1 and for carbinolamides in general.

Full text of DOI:10.1016/j.tetlet.2005.03.011

Tetrahedron
Letters
Tetrahedron Letters 46 (2005) 3079–3083
General-buffer catalysis of the reaction of
N-(hydroxymethyl)benzamide: a new pathway for the
aqueous reaction of carbinolamides
Amanda G. Mennenga, Amy L. Johnson and Richard W. Nagorski*
Department of Chemistry, Illinois State University, Normal, IL 61790-4160, USA
Received 4 January 2005; revised 1 March 2005; accepted 1 March 2005
Available online 18 March 2005
Abstract—The second-order rate constants for the general-buffer catalyzed breakdown of N-(hydroxymethyl)benzamide (1) in water
at 25 ꢀC, I = 1.0 (KCl) by pivalic, acetic, chloroacetic, and dichloroacetic acid were determined by initial rates. The observed rate
increased with increasing amounts of the acidic form of the buffer and a Brønsted correlation of a = 0.35 was determined. The results
presented here, represent the first evidence for a general-buffer catalytic mechanism for the aqueous reaction of 1 and for carbinol-
amides in general.
ꢁ
2005 Elsevier Ltd. All rights reserved.
1
–3
4–6
Carbonyl derivatives, such as acetals, acylals, and
7
Carbinolamides are biological intermediates, which are
enzymatically generated and cleaved during the biosyn-
–9
hemiacetals
play important roles in both synthetic
2
2–27
and biological chemistry. Mechanistic studies of the
aqueous reaction and methods of catalyzing this reac-
tion have been crucial in providing both an understand-
ing of their intrinsic reactivity and insight into the
thesis of peptide hormones.
amides are known to react via both hydronium ion
In general, carbinol-
1
5–21
and hydroxide-catalyzed routes.
1
–5
biological systems in which they are utilized.
The
(a) Hydroxide-catalyzed mechanism: Overall, a specific-
base catalyzed mechanism, which involves deproton-
ation of the hydroxyl group followed by rate-limiting
breakdown of the alkoxide to form aldehyde and the
broad nature of the types of functionality found within
these structurally similar groups are reflected in the
diverse mechanisms by which they react. For example,
both acetals and acylals were thought to react exclu-
sively via specific-acid catalyzed mechanisms, however,
examples of acetals were found, which reacted by gen-
1
5–21
anionic amide (Scheme 1).
(b) Hydronium ion-catalyzed mechanism: This reaction is
thought to occur via specific-acid catalyzed protonation
20
1
0–14
eral-acid mechanisms.
In addition, acylals have
of the amide portion of the carbinolamide, followed by
rate-limiting breakdown to the amide and protonated
been proposed to react through ester hydrolysis mecha-
nisms under circumstances where the electron-donating
ability of the alkoxy group was diminished due to the
presence of electron-withdrawing groups. The focus
of the study described here was on the general-buffer
catalysis of the aqueous reaction of carbinolamides,
which were previously thought to react only via spe-
cific-acid catalysis in the acidic region of their pH-rate
1
6,18,20
aldehyde (Scheme 2).
must be included in the rate expression to explain the
(A water reaction (kHOH)
4
,5
16,18,20
appearance of the pH-rate profile, see Eq. 1.)
ꢁ
K
þ K
a
½HO ꢀ
þ
kobsd ¼ k ½H ꢀ þ k
þ k1 K
ð1Þ
H
HOH
ꢁ
w
a
½HO ꢀ
1
5–21
profiles.
The dichotomy between the currently accepted mecha-
nisms and the existence of an enzyme, which catalyzes
the breakdown of carbinolamides was puzzling when
the general-acid/general-base catalytic schemes, often
associated with enzymatic systems, was considered.
If the lyase portion of PAM (peptidylglycine a-amidating
Keywords: Carbinolamide; N-(Hydroxymethyl)benzamide; Buffer
catalysis.
2
8,29
*
Corresponding author. Tel.: +1 309 438 8978; fax: +1 309 438
538; e-mail: rnagor@ilstu.edu
5
0
040-4039/$ - see front matter ꢁ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2005.03.011

3080
A. G. Mennenga et al. / Tetrahedron Letters 46 (2005) 3079–3083
O
OH
O
O
O
Ka
H+
O
k'1
O
+
H+
+
R
N
H
R
N
H
R
NH
NH2
H
H
Scheme 1. Specific-base catalyzed reaction.
H
H
O
OH
H
O
OH
K'a
O
O
k'H
O
+ H+
R
N
H
+
R
N
H
R
NH
R
NH2
H
H
Scheme 2. Specific-acid catalyzed mechanism.
2
2–27
monooxygenase, EC 1.14.17.3)
utilizes a general-
concentration of chloroacetic acid (see Fig. 1) showed
a linear correlation between [buffer] and kobsd for the
base/acid catalytic method to cleave the carbinolamide
intermediates (such a mechanism has been sugges-
reaction of 1 in H O, at 25 ꢀC, I = 1.0 (KCl). From Fig-
2
2
6,30
ted),
why has general-catalysis of the aqueous reac-
ure 1, as the pH of the solution decreased and, therefore,
the concentration of the acidic form of the buffer rose,
there was an associated rise in the slope of the correl-
ation, yielding an initial conclusion that the acidic form
of the buffer was catalyzing the reaction of 1.
tion of carbinolamides not been previously observed?
Herein, we provide the first evidence for general-buffer
catalysis of the aqueous reaction of a carbinolamide
(
N-(hydroxymethyl)benzamide, 1), as a pathway for its
breakdown in the acidic and water-catalyzed regions
of its pH-rate profile.
The y-intercepts, of the plots in Figure 1, yielded the rate
of the buffer independent reaction at that pH, and those
values were in good agreement with previous studies of
the aqueous reaction of 1 (see Table 1). The second-
order rate constant for the reaction of 1 (kHA), as a func-
A
car
¼
ꢁkobsd
t
ð2Þ
ðAcar þ ð1:8ÞðAamideÞÞ
20
tion of the acidic form of the buffer, was obtained by
plotting the second-order rates (k ) from Figure 1
N-(Hydroxymethyl)benzamide (1) was synthesized and
2
0
0,31
purified as previously reported.
1
The conversion of
into benzamide (2) and formaldehyde was followed
HA
kobsd vs [total buffer]) versus the fraction of the acidic
(
form of the buffer and was found to be k
=
HA
in H O, I = 1.0 (KCl), at 25 ꢀC in the presence of vary-
2
ꢁ
7
ꢁ1 ꢁ1
2
.0 · 10
M s for chloroacetic acid (see Supple-
ing concentrations of pivalate, acetate, chloroacetate, or
dichloroacetate buffer, as a function of pH. A typical
experiment involved the injection of a volume of concen-
mental materials for plot). Similar studies to those
described above were performed with acetic acid, dichlo-
roacetic acid, and pivalic acid. In all cases, a linear cor-
relation between the [buffer] and kobsd was found and
y-intercepts correlated with the calculated values from
3
3
trated stock solution of 1, dissolved in CH CN, into a
3
thermally equilibrated aqueous solution, yielding a final
ꢁ
5
ꢁ4
20
substrate concentration of 5 · 10 –1 · 10 M. The
progress of the reactions were monitored by following
the disappearance of the starting material and the
2
0
previous studies (plots of kobsd vs [buffer] available in
the Supplemental material).
3
2
appearance of amide, using HPLC. As a result of the
2
sluggish nature of these reactions, an initial rates
0
Shown in Table 1 are the second-order rate constants
kHA), determined as a function of the acidic form of
2
8
(
method was employed where the disappearance of
the first 3–4% of 1 and the appearance of 2 was fol-
lowed. The observed rate of the reaction (kobsd) was
determined from the linear plot of the fraction of 1
remaining versus time (s) according to Eq 2, where Acar
was the observed area of the peak of 1, Aamide was the
area of the peak for the amide product and a correction
factor of 1.8 was used that statistically corrected the ob-
served area of 2 for differences in the molar absorptivi-
ties between 1 and 2, at the wavelength at which
the buffer, for the aqueous reaction of 1 at 25 ꢀC,
2.5
2
.0
pH = 2.1
1.5
1.0
0.5
0.0
pH = 2.5
3
2
reaction progress was followed.
pH = 3.8
Previous investigations, performed at low buffer concen-
trations ([B ] 6 0.05 M), did not report any buffer
tot
catalysis, leading to the conclusion that the acid-pro-
moted reaction occurred via a specific-acid catalyzed
0
.0
0.2
0.4
0.6
0.8
1.0
[
Chloroacetic acid]tot (M)
2
0
mechanism. Further investigations at higher buffer
concentrations have shown that as the [buffer] was in-
creased, at the same pH, there was an increase in kobsd
for the reaction of 1. A plot of kobsd versus the total
Figure 1. Effect of increasing concentrations of chloroacetic acid on
the rate of the aqueous reaction (kobsd) of N-(hydroxymethyl)benz-
amide in H O at 25 ꢀC, I = 1.0 (KCl) at pHÕs = 2.1, 2.5, and 3.8.
2

A. G. Mennenga et al. / Tetrahedron Letters 46 (2005) 3079–3083
3081
Table 1. Second-order rate constants for the buffer-catalyzed breakdown of N-(hydroxymethyl)benzamide (1) and the experimental and calculated
rates for the buffer-independent reaction in H
2
O, at 25 ꢀC, I = 1.0 (KCl)
a
b
ꢁ1 ꢁ1
e
ꢁ1
calc
f
Buffer
CH CO
pK
a
pH
k
HA (M
s
)
kbkg (s
)
kbkg
ꢁ
ꢁ9
9
ꢁ9
ꢁ9
ꢁ9
ꢁ9
ꢁ8
ꢁ8
ꢁ7
ꢁ9
3
2
H
4.76
2.86
5.61
(2.5 ± 0.2) · 10
(1.4 ± 0.6) · 10
(3.4 ± 0.4) · 10
(3.0 ± 0.4) · 10
(2.2 ± 0.1) · 10
(4.7 ± 0.3) · 10
(1.9 ± 0.2) · 10
(2.0 ± 0.3) · 10
3.2 · 10
2.1 · 10
2.9 · 10
2.9 · 10
2.4 · 10
6.0 · 10
2.3 · 10
2.1 · 10
ꢁ
8 c
7 c
4
3
.44
.80
(5.2 ± 0.5) · 10
ꢁ9
ꢁ9
ClCH CO
2
2
H
H
3.81
ꢁ
ꢁ8
2
2
.51
.10
(2.0 ± 0.3) · 10
ꢁ8
ꢁ
ꢁ
6 d
8 d
ꢁ7
ꢁ9
Cl
(CH
2
CHCO
2
1.29
5.05
1.50
4.92
(2.0 ± 0.1) · 10
(4.1 ± 0.3) · 10
3 3 2
) CCO H
a
b
c
d
e
f
pK values from Ref. 45b.
a
pH of the solution in which the experiment was performed.
0
Calculated by plotting k
for the buffer-catalyzed reaction versus fraction of the buffer in the acidic form.
HA
Calculated from the slope of the kobsd versus fraction acidic form of the buffer.
The y-intercept values of the plots of kobsd versus [buffer].
Calculated using Eq. 1 (see Ref. 20).
I = 1.0 (KCl). From Table 1, as the acidity of the buffer
increases, its catalytic effectiveness also increases. This
observation can be more easily observed in Figure 2
where the Brønsted correlation between the acidity of
H
O
H
O
O
O
kHA
+
HA
NH2
+
+ A
N
H
H
H
the buffer catalyst was plotted versus logkHA, and re-
1
3
2
+
sulted in an a = 0.35. A point for H O has also been in-
3
+
cluded in the plot (rate for H O -catalyzed reaction was
Scheme 3. General-acid catalyzed mechanism.
3
ꢁ6
ꢁ1 ꢁ1 20
determined previously, k = 7.3 · 10
M s ) and
H
showed a good correlation with the acetate buffers indi-
cating a consistent mechanism between the buffer-
catalyzed reactions and the hydronium-ion catalyzed
reaction. Even more interesting was the observation that
the second-order rate for the water-catalyzed reaction
reacts via the same mechanism as the acetate and hydro-
nium ion-catalyzed reactions.
The catalytic effect of carboxylate buffers can be inter-
36
preted in two ways:
ꢁ
11
ꢁ1 ꢁ1 20,34
(
kHOH = 3.1 · 10
M s ) of 1 correlated well
with the buffer and hydronium ion points. This observ-
ation suggested that the water-catalyzed reaction, a diffi-
(a) General-acid catalysis (Scheme 3): The rate determin-
ing step of the reaction involves proton transfer from an
3
5
cult reaction to study due to the slow reaction rates,
+
H source to the amide portion of 1 as breakdown into 2
and 3 occurs. In Scheme 3, proton transfer to the nitro-
gen has been shown as the carbinolamide cleaves, how-
ever an equally valid mechanism could be written
wherein proton transfer occurs at the carbonyl oxygen
-
-
-
-
-
-
4
5
6
7
8
9
37
of the amide. The mechanism (Scheme 3) correlates
with the suggested mechanisms for the general-acid cat-
alyzed breakdown of acetals and would lead to an over-
all rate expression for the reaction of carbinolamides
shown in Eq. 3.
(b) Specific-acid followed by general-base (Scheme 4):
Here, the amide portion of the carbinolamide is proton-
ated, followed by rate-limiting general-base catalyzed
deprotonation of the hydroxyl group to produce formal-
-
-
10
11
-
4
-2
0
2
4
6
pKa
8
10 12 14 16 18
20,37
dehyde and a tautomeric form of benzamide.
mechanism yields the same acid dependence as in
Such a
HA
Scheme 3 (see Eqs. 3 and 4, where K
is the K of
a
Figure 2. Brønsted plot of the logkHA (second-order rate of buffer
catalyzed reaction of N-(hydroxymethyl)benzamide in H
I = 1.0 (KCl)) versus the pK for the acid catalysts used to catalyze the
a
the buffer catalyst and k [A ] can be dropped from the
ꢁ
2
O at 25 ꢀC,
2
denominator of Eq. 4 as it is assumed to be slow com-
pared to k 1).
a
reaction.
ꢁ
H
H
O
O
OH
O
OH
k2
O
k1
+ H+
+
N
H
+ A
N
H
NH
H
H
k-1
1
+ HA
Scheme 4. Specific-acid followed by general-base.

3082
A. G. Mennenga et al. / Tetrahedron Letters 46 (2005) 3079–3083
Based on the results presented here, both of the mecha-
nisms discussed above are feasible. The general-acid cat-
alyzed mechanism could be criticized on the basis of the
generation of protonated formaldehyde (3). Mechanistic
studies of the reaction of acetals have shown that gen-
eral-acid catalysis only occurred in those cases where a
evidence that the aqueous reaction of carbinolamides can
occur via a general acid/base mechanism. This observa-
tion could provide insight into the mechanism by which
PAM catalyzes the breakdown of the carbinolamide
intermediates generated in peptide hormone synthesis.
1
–3
stable carbocation was created. Under the conditions
of the studies described here, 3 would not be stable en-
Acknowledgments
3
8–44
ough to exist in solution
and, therefore, the gen-
eral-acid mechanism suggested in Scheme 3 would not
be possible. However, the mobility of the proton on 3
versus the alkylated version in acetal studies could avoid
This work was supported by an award from Research
Corporation and the donors to the Petroleum Research
Fund, administered by the American Chemical Society.
4
5
the generation of this unstable intermediate.
þ
k
obsd ¼ k
H
½H ꢀ þ kHA½HAꢀ þ kHOH
Supplementary data
ꢁ
K ½HO ꢀ
a
þ kHO
ð3Þ
ð4Þ
ꢁ
The plots of the change in the kobsd versus the change in
K þ K ½HO ꢀ
w
a
[
buffer] for acetic acid, dichloroacetic acid, and pivalic
plotted versus the fraction
0
acid. Also plots of the k
HA
þ
ꢁ
HA
k k ½H ꢀ½A ꢀ k k K ½HAꢀ
of buffer in the acidic form are available free of charge
via the Internet. Supplementary data associated with
this article can be found in the online version, at
doi:10.1016/j.tetlet.2005.03.011.
1
2
1
2
a
k
HA½HAꢀ ¼ k
¼ k
ꢁ
ꢁ
þ k ½A ꢀ
þ k ½A ꢀ
ꢁ
1
2
ꢁ1
2
HA
k k K ½HAꢀ
1
2
a
¼
k
ꢁ
1
References and notes
Further evidence supporting a mechanism such as that
shown in Scheme 4 comes from buffer catalysis studies
of the aqueous reaction of hemiacetals formed from
1
2
. Dunn, B.; Bruice, T. C. Adv. Enzymol. 1973, 37, 1.
. Cordes, E. H.; Bull, H. G. Chem. Rev. 1974, 74, 581–603.
3. Fife, T. H. Acc. Chem. Res. 1972, 5, 264–272.
. Fife, T. H. Adv. Phys. Org. Chem. 1975, 11, 108.
formaldehyde and a variety of alcohols (pK Õs ranging
a
8
from 12.4–16). Cross-interaction coefficients generated
4
by correlating structure/reactivity studies and Brønsted
aÕs (a-values varied between 0.28 and 0.36 for formalde-
hyde hemiacetals) led to the conclusion that, under
acidic conditions, the hemiacetals studied reacted via a
buffer-catalyzed mechanism similar to that shown in
5. Capon, B. Chem. Rev. 1969, 69, 407.
6. Hall, C. D.; Le, V. T. J. Chem. Soc., Perkin Trans. 2 1998,
1483–1488.
7
. Engall, K. M.; McClelland, R. A.; Sørensen, P. E. Can.
J. Chem. 1999, 77, 978–989.
8
8. Funderburk, L. H.; Aldwin, L.; Jencks, W. P. J. Am.
Chem. Soc. 1978, 100, 5444–5459.
Scheme 4. The structural relationship of these hemiace-
tals to 1 and the similarity of the Brønsted a-values to
that determined for 1 provide support for the reaction
of 1 occurring via mechanism like that shown in Scheme
9
. Sørensen, P. E.; Jencks, W. P. J. Am. Chem. Soc. 1987,
09, 4675–4690.
1
1
1
0. Fife, T. H.; Jao, L. K. J. Am. Chem. Soc. 1968, 90, 4081.
1. Anderson, E.; Capon, B. J. Chem. Soc. B 1969, 1033.
4
. Although no conclusion, with respect to the mecha-
nism of the catalysis by carboxylate buffers, is possible
based on the results presented here, cross-interaction
coefficient studies should lead to a resolution of this
12. Anderson, E.; Fife, T. H. J. Am. Chem. Soc. 1969, 91,
7163–7166.
13. Fife, T. H.; Brod, L. H. J. Am. Chem. Soc. 1970, 92, 1681.
7
–9
1
4. Capon, B.; Nimmo, K. J. Chem. Soc., Perkin Trans. 2
975, 1113.
kinetic ambiguity.
1
1
5. Bundgaard, H. In Design of Prodrugs; Bundgaard, H.,
Ed.; Elsevier: Amsterdam, 1985; pp 1–92.
6. Bundgaard, H.; Buur, A. Int. J. Pharm. 1987, 37, 185–194.
7. Bundgaard, H.; Johansen, M. Int. J. Pharm. 1980, 5, 67–
In conclusion, it has been shown that carboxylate
buffers catalyze the aqueous reaction of N-(hydroxy-
1
1
1
5–21
methyl)benzamide (1). Previous studies, focused
on the aqueous breakdown of carbinolamides, failed
to note any catalytic effect by buffers, which, for 1,
7
8. Bundgaard, H.; Johansen, M. Int. J. Pharm. 1984, 22, 45–
7.
1
2
0
was probably due to the low total [buffer] used in those
studies and the modest rate enhancements observed for
the catalysts in general. It can be argued that the cata-
lytic effect of acetate derivatives are as a result of
general-acid catalysis or specific-acid followed by general-
base catalysis with further studies necessary to resolve
this issue. Also, the correlation of the second-order rate
constants in the Brønsted plot suggests that the buffer,
hydronium ion, and water-catalyzed reactions occur
via a consistent mechanism, which is quite remarkable
56.
19. Johansen, M.; Bundgaard, H. Arch. Pharm. Chem. Sci.
Edn. 1979, 7, 175–192.
2
0. Tenn, W. J.; French, N. L.; Nagorski, R. W. Org. Lett.
001, 3, 75–78.
2
2
2
1. Zaugg, H. E.; Martin, W. B. Org. React. 1965, 14, 52–269.
2. Kulathila, R.; Merkler, K. A.; Merkler, D. J. Nat. Prod.
Rep. 1999, 16, 16.
2
3. Eipper, B. A.; Perkins, S. N.; Husten, E. J.; Johnson, R.
C.; Keutmann, H. T.; Mains, R. E. J. Biol. Chem. 1991,
2
66, 7827–7833.
when the broad range in their pK Õs are considered.
The results of these studies have provided the first
24. Eipper, B. A.; Stoffers, D. A.; Mains, R. E. Annu. Rev.
Neurosci. 1992, 15, 57–85.
a

A. G. Mennenga et al. / Tetrahedron Letters 46 (2005) 3079–3083
3083
2
2
5. Eipper, B. A.; Mains, R. E. Annu. Rev. Physiol. 1988, 50,
33–344.
6. Kolhekar, A. S.; Bell, J.; Shiozaki, E. N.; Jin, L. X.;
Keutmann, H. T.; Hand, T. A.; Mains, R. E.; Eipper, B. A
Biochemistry 2002, 41, 12384–12394.
34. Second-order rate for the water reaction was calculated by
dividing pseudo-first rate (see Ref. 20) by the concentra-
3
ꢁ9
ꢁ1
tion of water (1.7 · 10
s /55.5 M).
35. The half-life of the water-catalyzed reaction is ꢂ13 years.
36. Lowry, T. H.; Richardson, K. S. Mechanism and Theory in
Organic Chemistry; Harper & Row: New York, 1987.
37. Arguments for protonation on oxygen or nitrogen could
be made based on acid-catalyzed amide hydrolysis studies
(see Ref. 20).
2
7. Bell, J.; Ash, D. E.; Snyder, L. M.; Kulathila, R.;
Blackburn, N. J.; Merkler, D. J. Biochemistry 1997, 36,
1
6239–16246.
2
2
3
8. Jencks, W. P. Catalysis in Chemistry an Enzymology;
Dover Publications: Mineola, NY, 1975.
38. Young, P. R.; Jencks, W. P. J. Am. Chem. Soc. 1977, 99,
8238–8248.
39. Craze, G.-A.; Kirby, A. J.; Osborne, R. J. Chem. Soc.,
Perkin Trans. 2 1978, 357–368.
40. Knier, B. L.; Jencks, W. P. J. Am. Chem. Soc. 1980, 102,
6789–6798.
9. Fersht, A. Enzyme Structure and Mechanism, 2nd ed.;
W.H. Freeman & Company: New York, 1985.
0. Mounier, C. E.; Shi, J.; Sirimanne, S. R.; Chen, B. H.;
Moore, A. B.; Gillwoznichak, M. M.; Ping, D. S.; May,
S. W. J. Biol. Chem. 1997, 272, 5016–5023.
3
1. Nair, B. R.; Francis, J. D. J. Chromatogr. 1980, 195, 158–
41. Jencks, W. P. Acc. Chem. Res. 1980, 13, 161–169.
42. Jencks, W. P. Chem. Soc. Rev. 1981, 10, 345–375.
43. Amyes, T. L.; Jencks, W. P. J. Am. Chem. Soc. 1988, 110,
3677–3679.
1
61.
2. Solvent system used to separate 1 from 2 via HPLC was
0% 70/30 water/methanol and 10% methanol. The
3
9
chromatograms were generated by following the reactions
at 254 nm.
44. Amyes, T. L.; Jencks, W. P. J. Am. Chem. Soc. 1989, 111,
7900–7909.
3
3. Acetic acid studies were performed at three different pHÕs
while those performed in the presence of dichloro-
acetic acid and pivalic acid were investigated at only one
pH.
45. (a) The pKa of protonated formaldehyde should be
ꢂꢁ1.74; (b) CRC Handbook of Biochemistry and Molec-
ular Biology; Fasman, G. D., Ed., 3rd ed.; CRC: Cleve-
land, USA, 1976.