Isolation and characterization of a novel tetrahydro-[2,2′]bipyrrolyl dimer as an impurity from a Knorr reaction

Article abstract of DOI:10.1002/jhet.5570410521

A novel dimer, tetraethyl 2,2′,3,3′-tetramethyl-1,1′,2, 2′-tetrahydro-4H,4′H-2,2′-bipyrrolyl-5,5,5′, 5′-tetracarboxylate, has been isolated as an impurity (0.4% yield) from a Knorr reaction for the synthesis of ethyl 3,5-dimethylpyrrole-2-carboxylate from 2,4-pentanedione and diethyl oximinomalonate in a dissolving zinc reduction. The solid-state structure of this novel dimer was determined by X-ray crystallography. Knorr reactions typically rely upon the requisite pyrrole being the only water-insoluble crystalline material present in the reaction mixture, and so work-up and purification procedures for Knorr reactions should be monitored carefully given the water-insolubility of this dimer. Investigations regarding mechanistic implications and reductive dimerization are underway.

Full text of DOI:10.1002/jhet.5570410521

Sep-Oct 2004
777
Isolation and Characterization of a Novel Tetrahydro-[2,2']bipyrrolyl
Dimer as an Impurity from a Knorr Reaction
Alison Thompson*, Yousef Alattar, Cory S. Beshara, Rodney K. Burley,
T. Stanley Cameron and Katherine N. Robertson
Department of Chemistry, Dalhousie University, Halifax, Nova Scotia, B3H 4J3, CANADA
Received March 15, 2004
A novel dimer, tetraethyl 2,2',3,3'-tetramethyl-1,1',2,2'-tetrahydro-4H,4'H-2,2'-bipyrrolyl-5,5,5',5'-
tetracarboxylate, has been isolated as an impurity (0.4% yield) from a Knorr reaction for the synthesis of
ethyl 3,5-dimethylpyrrole-2-carboxylate from 2,4-pentanedione and diethyl oximinomalonate in a dissolv-
ing zinc reduction. The solid-state structure of this novel dimer was determined by X-ray crystallography.
Knorr reactions typically rely upon the requisite pyrrole being the only water-insoluble crystalline material
present in the reaction mixture, and so work-up and purification procedures for Knorr reactions should be
monitored carefully given the water-insolubility of this dimer. Investigations regarding mechanistic implica-
tions and reductive dimerization are underway.
J. Heterocyclic Chem., 41, 777 (2004).
Introduction.
Results and Discussion.
The Knorr [1] reaction and derivatives thereof [2-4] rep-
resent the most efficient and economical synthetic method-
ology for the preparation of a wide variety of functional-
ized pyrroles. This procedure uses readily available 1,3-
diketones, and oximes of either acetoesters or malonates, to
give pyrroles functionalized with a variety of substituents
via a dissolving zinc reduction. The basic mechanism of
the Knorr reaction has long been established [2,5,6], and
yields are reliably in the range 30-60%, with perhaps 45%
being considered average [4]. The starting materials are
generally extremely cheap and the procedure is well suited
for large-scale reactions (1-5 mole scales are routinely
manageable in a laboratory environment).
Scheme 1 shows the Kleinspehn variation [7] of the
Knorr reaction for the preparation of ethyl 3,5-dimethyl-
pyrrole-2-carboxylate (1), whereby diethyl oximino-
malonate (2) is reacted with 2,4-pentanedione (3) in the
presence of zinc. This variation, involving the oximes of
malonate esters, is often called the Fischer-Fink reaction as
Fischer and Fink isolated a small amount of ethyl 3,5-
dimethylpyrrole-2-carboxylate from the Knorr reaction of
2,4-pentanedione and α-oximinoacetoacetate under dis-
solving zinc reduction conditions [8]. The reaction is con-
ducted by first preparing the diethyl oximinomalonate by
nitrosation of diethyl malonate with an aqueous solution of
sodium nitrite. Nitrogen oxides are evolved during this

778
A. Thompson, Y. Alattar, C. S. Beshara, R. K. Burley, T. S. Cameron and K. N. Robertson
Vol. 41
nitrosation, and so the reaction should be conducted in an
efficient fume-hood. The crude oxime (2) is then added
drop-wise to a vigourously stirring solution of 2,4-pen-
tanedione in acetic acid, with concomitant slow addition of
zinc dust. As such, the diethyl oximinomalonate is reduced
to give diethyl aminomalonate (4) in-situ, which immedi-
ately reacts with the diketone (3); self-condensation of 4 is
thus minimized. With slow addition of both the oxime and
the zinc dust, the temperature of the exothermic reaction is
maintained at 60-80 °C, a temperature range that gives
coloured, glassy and angular. Pyrroles are often isolated as
coloured precipitates that are >98% pure by NMR spectro-
scopic analysis, and so the visual appearance of the crys-
tals did not induce immediate concern or interest.
1
However, the H NMR spectrum of the salmon-coloured
crystals indicated the absence of a diagnostic pyrrole N-H
signal (8-10 ppm). Instead, an amino N-H signal was
observed (2.73 ppm), along with two ethyl ester signals.
The presence of two ethyl esters in unique environments
was confirmed by 13C NMR spectroscopy; two electron-
maximized yields. Zinc acetate often precipitates as the
reaction progresses, making stirring more difficult,
although the yields are not generally affected [4]. Sodium
(or ammonium) acetate (2.0 molar equivalents) is often
added to the reaction mixture in order to keep the zinc
acetate in solution and hence facilitate processing.
deficient alkene-type carbon environments were also evi-
dent (signals at 132.36 and 135.95 ppm). Mass spectrome-
try revealed a base peak of 240 (m/z) for the isolated impu-
rity, and closer inspection revealed a molecular ion of low
intensity at 480 (m/z). This led us to deduce the structure
of the impurity to be that of the dimer 8 (Scheme 2).
Attempted cleavage of the central C-C bond under thermal
conditions, external to the mass spectrometer, was unsuc-
cessful, and we were unable to convert 8 into 1. X-ray
crystallographic analysis [10] confirmed the dimeric struc-
ture of 8 (Figure 1). Selected bond lengths and angles are
given in Table 1.
The work-up procedure for Knorr reactions is straight-
forward as the pyrrole is usually the only water-insoluble
solid present in the reaction mixture [4,9]. Indeed, the
work-up procedure merely involves adding 2-3 volumes of
iced water to the crude reaction mixture once it has been
carefully decanted away from residual zinc solid, which is
potentially pyrophoric [4]. The precipitated crude pyrrole
is then isolated by filtration, dissolved in a solvent such as
dichloromethane and the layers are then separated; this
removes the large amount of water trapped within the
initial cake. The pure pyrrole product is then obtained by
crystallization. The key purification step within this proce-
dure involves separation, by filtration, of the crude precip-
itate and the aqueous acetic acid solution after addition of
iced water to the reaction mixture. The success of this
purification relies on the fact that the required pyrrole is
the only water-insoluble crystalline material present [4,9].
Following the Knorr procedure described above, we
obtained a reasonable yield (44%) of ethyl 3,5-dimethyl-
pyrrole-2-carboxylate (1) on a 1.0 mole scale, after two
crystallizations from dichloromethane and hexane. Upon
the third crystallization, it was noted that the crystals were
different in appearance to those of the previous two crops;
pyrrole 1 was obtained as fine white platelets in crops one
and two, whilst the third crop of crystals was pale salmon-
Analysis of the crystal structure shows that only one half
of the dimeric structure is crystallographically unique.
Within the five-membered ring, one of the bonds, C(2)-
Figure 1: ORTEP diagram of 8; thermal ellipsoids drawn at the 50%
probability level. Hydrogen atoms, apart from H(1) have been omitted for
clarity. Only half of the dimer is crystallographically unique.

Sep-Oct 2004
Isolation and Characterization of a Novel Tetrahydro-[2,2']bipyrrolyl Dimer
779
Table 1
Selected Bond Lengths (Å) and Angles (°) for 8
Bond lengths
O(1)-C(5)
O(2)-C(6)
O(4)-C(8)
N(1)-C(1)
N(1)-H(1)
C(1)-C(5)
C(2)-C(3)
C(3)-C(4)
C(4)-C(12)
C(6)-C(7)
1.196(4)
1.458(4)
1.324(4)
1.446(4)
0.84(3)
1.532(4)
1.321(4)
1.498(4)
1.530(4)
1.488(6)
O(2)-C(5)
O(3)-C(8)
O(4)-C(9)
N(1)-C(4)
C(1)-C(2)
C(1)-C(8)
C(2)-C(11)
C(4)-C(4)#1
C(9)-C(10)
1.319(4)
1.198(4)
1.470(4)
1.497(3)
1.514(4)
1.550(4)
1.501(4)
1.563(6)
1.467(6)
Figure 3: Hydrogen bonding within crystal packing of 8
Bond angles
N(1)-C(1)-C(5)
N(1)-C(1)-C(8)
N(1)-C(1)-C(2)
N(1)-C(4)-C(3)
N(1)-C(4)-C(4)#1 108.7(3)
N(1)-C(4)-C(12) 111.2(2)
C(1)-N(1)-C(4)
C(1)-N(1)-H(1)
C(4)-N(1)-H(1)
C(2)-C(1)-C(5)
C(5)-O(2)-C(6)
C(8)-O(4)-C(9)
C(2)-C(1)-C(8)
C(5)-C(1)-C(8)
C(3)-C(2)-C(11)
107.0(2)
C(3)-C(2)-C(1)
C(11)-C(2)-C(1)
C(2)-C(3)-C(4)
C(3)-C(4)-C(12)
C(3)-C(4)-C(4)#1
108.3(2)
112.1(2)
104.9(2)
101.4(2)
124.2(3)
113.9(3)
112.6(2)
110.3(2)
Å], and the two rings adopt a twisted geometry relative to
one another, with an angle of 61.22(8)° between the two
ring planes. In the solid-state structure, long chains of mol-
ecules are aligned parallel to the c axis (Figure 2), and
there are N-H…O hydrogen bonds between the molecules
in adjacent chains (Figure 3). Indeed, the H(N) hydrogen
atom adopts a bifurcated hydrogen bonding arrangement.
There is the N(1)-H(1)…O(3') hydrogen bond that exists
between the chains, as mentioned above [N(1)-H(1)
0.84(3) Å, H(1)…O(3') 2.43(3) Å, N(1)…O(3') 3.218(3)
Å, and N(1)-H(1)…O(3') 156(3)°; the symmetry required
to generate O(3') is 1-x, -y, 1-z]. The second component of
the hydrogen bond closes an internal five-membered ring
involving N(1), H(1), O(3), C(8) and C(1) all in the same
monomer unit, [N(1)-H(1) 0.84(3) Å, H(1)…O(3) 2.41(3)
Å, N(1)…O(3) 2.797(3) Å, and N(1)-H(1)…O(3)
109(3)°].
C(12)-C(4)-C(4)#1 112.1(2)
109.6(2)
110(2)
115(2)
116.2(3)
116.2(3)
116.8(3)
110.8(2)
106.0(2)
127.5(3)
O(1)-C(5)-O(2)
O(1)-C(5)-C(1)
O(2)-C(5)-C(1)
O(2)-C(6)-C(7)
O(3)-C(8)-O(4)
O(3)-C(8)-C(1)
O(4)-C(8)-C(1)
C(10)-C(9)-O(4)
124.3(3)
123.6(3)
111.9(3)
106.5(3)
124.2(3)
124.7(3)
111.0(3)
108.2(3)
Symmetry transformations used to generate equivalent atoms: #1 -
x+1,y,-z+3/2.
The interesting observation that the dimer was isolated
as the D,L-form, rather than the meso-form, is indicative of
high diastereoselectivity in the dimerization process. An
electronic search of Chemical Abstracts revealed no hits
for the basic skeleton of compound 8, and only a handful
of reported 2,2'-bipyrrolidine [11-13] systems (reduced
skeleton c.f. 8) with C-substitution at the 2 and 2' positions
accompanied by bis-substitution at the 5 and 5' positions.
As such, we believe that dimer 8 represents a completely
novel structural skeleton.
Analysis of the mechanism for the Knorr reaction [4],
reveals a number of intermediates, as shown in Scheme 1.
The pyrrole is formed via nucleophilic attack on interme-
diate 7, most probably by generated (and added, depending
upon the experimental protocol) acetate, thus resulting in
expulsion of the 5-membered ring that immediately
achieves aromaticity and protonation of the anion on nitro-
gen. Presumably, compound 8 is formed via a reductive
dimerization of an intermediate within the Knorr reaction
(Scheme 2). It is likely that water is eliminated from the
Figure 2: Crystal packing of 8 viewed down the b axis. Hydrogen atoms
have been omitted for clarity.
C(3) [1.321(4) Å], is significantly shorter than the others,
confirming its assignment as a double bond. The ring itself
is quite planar; the root mean square deviation of the atoms
from the ring plane is 0.0594 Å. The two halves of the
dimer are joined by the C(4)-C(4') single bond [1.563(6)

780
A. Thompson, Y. Alattar, C. S. Beshara, R. K. Burley, T. S. Cameron and K. N. Robertson
Vol. 41
charge (m/z) for all values between 50 and the molecular ion, if
over 10% of the base peak. Intensities are reported in parentheses
as a percentage of the base peak. Melting points were collected
on a hot-stage apparatus and are uncorrected. All reagents and
solvents were used as received. Nitrogen oxides are evolved dur-
ing the nitrosation, and so the reaction should be conducted in an
efficient fume-hood. Due to its potentially pyrophoric nature the
powdered zinc residue is best discarded by dissolving in dilute
sulfuric acid, in a fume-hood [4].
cyclized intermediate 6 such as to give the conjugated
intermediate 7, and that reductive dimerization of 7 (or its
protonated form), under buffered dissolving zinc condi-
tions, gives the isolated dimer 8 via a radical anion.
Presumably, the reduction is faster than nucleophilic attack
in this case, and hence dimer is formed rather than pyrrole.
This may have been due to the presence of a 'micro-envi-
ronment' within the reaction mixture as, despite vigorous
stirring, clumping of the zinc tends to occur.
Reaction conditions were then varied whereby different
stoichiometries of the reductant zinc (0.5-5.0 eq.) were
used, and the rates of stirring were varied whilst keeping
within safe limits, given the large exotherm of the reaction.
The stoichiometry of sodium acetate was also varied (0-5.0
eq.), as reductive dimerization must compete with nucle-
ophilic attack by acetate on 7. Under these conditions,
TLC analysis of the reaction mixture often and variably
showed traces of the dimer, but we were unable to isolate
any more material, no matter what scale we performed the
Knorr reactions on (0.2-5.0 mol). Yields of 1 remained in
the region of 40-50%, whatever the reaction conditions, as
reported by others [4]. We are continuing to pursue the
efficient synthesis of compound 8 since its skeleton is
entirely unique, and these investigations will reveal more
about the mechanism of the Knorr reaction itself.
In conclusion, a water-insoluble dimer with a novel
tetrahydro-[2,2']bipyrrolyl-5,5,5',5'-tetracarboxylic acid
skeleton has been isolated as an impurity from a Knorr
reaction whilst synthesizing ethyl 3,5-dimethylpyrrole-2-
carboxylate. Given the propensity of the Knorr reaction, it
is remarkable that this dimer has not been previously iso-
lated. The tetraethyl 2,2',3,3'-tetramethyl-1,1',2,2'-tetrahy-
dro-4H, 4 'H-2,2'-bipyrrolyl-5,5,5',5'-tetracarboxylate (8)
must arise from reductive dimerization of intermediates in
the Knorr reaction. Given the reliance upon pyrroles being
the only water-insoluble product of the Knorr reaction, the
work-up and purification procedures for Knorr reactions
should be monitored carefully, as it is likely that small
amounts of dimers of this type are formed frequently.
Further studies are underway in order to determine the fac-
tors influencing the formation of the dimer.
2,2',3,3'-Tetramethyl-1,1',2,2'-tetrahydro-4H, 4 'H-2, 2'-bipyrrolyl-
5,5,5',5'-tetracarboxylate.
According to the general procedure for the synthesis of pyrroles
via the Knorr reaction [4], a saturated aqueous solution of NaNO
2
(207 g, 3.0 mol) was added drop-wise to a solution of diethyl mal-
onate (152 mL, 1.0 mol) in acetic acid (200 mL). The solution was
stirred overnight and the formed oil (crude oxime) separated. Water
(500 mL) was added to the acetic acid solution, which was then
extracted with diethyl ether (2 x 50 mL). The combined ethereal
fractions were carefully concentrated in vacuo to give a further
quantity of crude oxime. The two portions of oxime were combined
and then added drop-wise to a solution of 2,4-pentanedione (103
mL, 1.0 mol) and NaOAc (272 g, 2.0 mol) in acetic acid (500 mL),
in a 4-necked flask equipped with an overhead stirrer, a dropping
funnel, a powder funnel and a thermometer. During the addition of
the oxime, zinc dust (195 g, 3.0 mol) was added gradually to the
reaction mixture. The internal temperature was maintained at 65-80
°C with the aid of first a bath of hot water and then an ice-bath.
After the addition was complete, the reaction mixture was stirred
for a further hour and then the solution was decanted away from the
solid zinc residue before iced water (2.0 L) was added. After stand-
ing overnight, the off-white precipitate was separated by filtration
and then dissolved in CH Cl to enable removal of the water
2
trapped within the cake. The CH Cl was then replaced with
2
2
2
hexane, with the aid of a rotary evaporator, to effect crystalliza-
tion/precipitation. First and second crops of crystals yielded 68.3 g
(41%) and 5.1 g (3%), respectively, of ethyl 3,5-dimethylpyrrole-2-
carboxylate (1) as a white powder [14,15], R (1:4 EtOAc:hexane)
f
0.48. A further crystallization of the mother liquor at 4°C overnight
gave the title compound (8) as angular salmon-coloured glassy
crystals (0.9 g, 0.4 %), m.p. 94-97 °C; R (1:4 EtOAc:hexane) 0.40;
f
1
H NMR: δ 1.16 (6H, s, CH ), 1.20-1.27 (12H, m, 4 x CH CH ),
H
3
2
3
1.81 (6H, s, CH ), 2.73 (2H, br s, NH), 4.10-4.25 (8H, m, 4 x
3
CH CH ), 5.54 (2H, s, CH); C NMR: δ 13.64 (CH CH ), 13.93
13
2
3
C
2
(CH ), 13.95 (CH CH ), 23.31 (CH ), 60.98 (CH CH ), 61.65
3
3
(CH CH ), 73.52 (NC(CH )), 80.01 (C(CO Et) ), 132.36
2
3
3
2
3
2
3
3
2
2
(CC(CO Et) ), 135.95 (CH), 169.72 (CO Et), 171.09 (CO Et); m / z
2
2
2
2
+
+
480.2 (M , 5%), 240.1 (M /2, 100), 168.1 (11), 122.0 (11);
C H NO calc. 240.1236, actual 240.1287.
EXPERIMENTAL
12 18
4
X-ray Structure Determination.
General Procedures.
Crystallographic data and details of the refinement are given in
Table 1 [10]. All measurements were made on a Rigaku AFC5R
diffractometer equipped with graphite monochromated Mo Kα
radiation and a rotating anode generator. The data were collected
at a temperature of –60 °C using the ω-2θ scan technique to a
maximum 2θ value of 60.1°. Cell constants and an orientation
matrix for data collection were obtained from a least-squares
refinement using the setting angles of 25 reflections in the range
38.52 < 2θ < 39.99° and corresponded to a C-centered mono-
clinic cell. Based on the systematic absences, packing considera-
Proton and carbon nuclear magnetic resonance (NMR) spectra
were recorded at 250 and 63 MHz, respectively, in CDCl on an
3
AC 250 F NMR spectrometer in the Atlantic Region Magnetic
Resonance Centre at Dalhousie University. Chemical shifts (δ
and δ ) are quoted in parts per million (ppm), referenced to the
H
C
appropriate residual solvent peak, and all coupling constants are
reported in hertz (Hz). Mass spectra (m/z) were obtained using a
+
CEC 21-110B instrument (EI , 124°C probe temperature, 70.0
eV energy). Mass spectra are reported in units of mass over

Sep-Oct 2004
Isolation and Characterization of a Novel Tetrahydro-[2,2']bipyrrolyl Dimer
781
[1] L. Knorr, Justus Liebigs Ann. Chem. 236, 290 (1886).
tions, a statistical analysis of intensity distribution, and the suc-
cessful solution and refinement of the structure, the space group
was determined to be C2/c (#15). The data were corrected for
Lorentz and polarization effects. Data reduction was carried out
using the teXsan software package [16]. The structure was solved
by direct methods (SIR-92) [17] and expanded using Fourier tech-
[2] V. F. Ferreira, M. C. B. V. de Souza, A. C. Cunha, L. O. R.
Pereira, and M. L. G. Ferreira, Org. Prep and Proc. Int. 3 3, 411
(2001).
[3] A. R. Jones and G. P. Bean, Eds, "The Chemistry of
Pyrroles." Academic Press, London, 1977.
[4] J. B. Paine III, in "The Porphyrins" (D. Dolphin, ed.), Vol.
I, Chapter 4. Academic Press, 1978.
[5] J. W. Harbuck and H. Rapoport, J. Org. Chem. 36, 853
(1971).
[6] J. B. Paine III, J. R. Brough, K. K. Buller and E. E.
Erikson, J. Org. Chem. 52, 3986 (1987).
[7] G. G. Kleinspehn, J. Am. Chem. Soc. 77, 1546 (1955).
[8] H. Fischer and E. Fink, Hoppe-Seyler's Z. Physiol. Chem.
280, 123 (1944).
2
niques. Refinement was carried out on F data using SHELXL-97
[18]. The non-hydrogen atoms were refined anisotropically. The
position of H(N) was located in the Fourier difference map after
refinement of the heavy atoms; it was then allowed to refine with
an isotropic temperature factor equal to 1.20 times the isotropic
temperature factor of the bonded nitrogen atom. The remaining
hydrogen atoms were placed in geometrically calculated posi-
tions. They were allowed to ride on the heavy atom to which they
were bonded with isotropic temperature factors equal to 1.20
times the isotropic temperature of the heavy atom (1.50 times for
methyl hydrogens). Neutral atom scattering factors and anom-
alous dispersion corrections were taken from the International
Tables for X-ray crystallography [19].
[9] A. R. Butler and S. D. George, Tetrahedron 49, 7017
(1993).
[10] CCDC 230389 contains the supplementary data for this
p a p e r. These data may be obtained, free of charge, via
w w w.ccdc.cam.ac.uk/retrieving.html (or from the Cambridge
Crystallographic data Centre, 12 Union Road, Cambridge, UK; fax
+44 1223 336033; or deposit.ccdc.ca.ac.uk).
[11] J. B. Bapat and D. S. C. Black, J. Chem. Soc., Chem.
Commun., 902 (1966).
[12] J. B. Bapat and D. S. C. Black, Aust. J. Chem. 21, 2497
(1968).
[13] M. Cariou, R. Hazard, M. Jubault and A. Ta l l e c ,
Tetrahedron Lett. 22, 3961 (1981).
[14] H. Fischer and B. Walach, Chem. Ber. 58, 2818 (1925).
[15] J. B. Paine III and D. Dolphin, J. Org. Chem. 50, 5598
(1985).
Crystal data for 8, C
H N O , 480.56
24 36 2 8
Colourless needle, monoclinic, space group C2 /c (no. 15), a =
14.993(5) Å, b = 13.789(4) Å, c = 13.587(3) Å, α = 90°, β =
3
107.20(2)°, γ = 90°, V = 2683(1) Å , Z = 4, T = 213(2) K, 2θ
=
max
60.1°, Rigaku AFC5R diffractometer, Mo Kα (λ = 0.7107 Å),
reflections collected = 4234, independent reflections = 3928 (R
= 0.0419), final R ([I>2s(I)] R1 = 0.0679, wR2 = 0.2002), R
int
2
indices (all data [R1 = 0.2046, wR2 = 0.2488]), gof (F ) = 1.490.
Acknowledgements.
We thank the Natural Sciences and Engineering Research
Council of Canada (NSERC) and Dalhousie University for finan-
cial support.
[16] Molecular Structure Corporation. (1997-1999). teXsan for
Windows. Single Crystal Structure Analysis Software. Version 1.06.
MSC, 9009 New Trails Drive, The Woodlands, TX 77381, USA.
[17] A. Altomare, M. G. Cascarano, C. Giacovazzo and A.
Guagliardi, J. Appl. Cryst. 26, 343 (1993).
REFERENCES AND NOTES
[18] Sheldrick, G.M. (1997). SHELXL-97. University of
Goettingen, Germany..
*
Corresponding author: Alison Thompson; E-mail:
[19] V. C. International Tables for Crystallography, Ed. A.J.C.
Wilson, Kluwer Academic Publishers, Dordrecht: Tables 6.1.1.4 (pp.
500-502), 4.2.6.8 (pp. 219-222) and 4.2.4.2 (pp. 193-199).
alison.thompson@dal.ca; Fax: (1) 902 494 1310; Tel: (1) 902 494
3305.