Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
1
A new approach in rock-typing, documented by a case study of layer-cake
reservoirs in field "A", offshore Abu Dhabi (U.A.E.)
Bruno GRANIER1
Abstract: In carbonate reservoirs, the relationship between porosity Ø, a measure of the combined
volumes of several kinds of pore space ( e.g. interparticle and separate-vug), and permeability K are
neither linear nor logarithmic, hence only weakly correlatable. Approaches to an estimation of
permeability that employ both petrographical and petrophysical parameters, the so-called rock-
typing techniques, have proven to be the most nearly precise. However in many studies simple K/Ø
cross-plots are used for each rock-type to provide trendlines from which K values are derived as a
function of Ø values; this is common practice even though the coefficient of correlation r 2 departs
significantly from 1.
This paper describes and provides examples of an improved technique of rock-type classification in
which each rock-type is characterized by a discrete and unique Gaussian distribution of log K. It
facilitates upscaling for it suggests the use of a single geometric mean value for permeability and a
corresponding standard deviation (variance, or coefficient of variation) for each rock-type.
This new technique can be used in uncored wells by extrapolating these determinations into wireline
logs as documented below in a case study of layer-cake reservoirs in a field of the Abu Dhabi
offshore (U.A.E.).
Key Words: Petrophysics; sedimentary petrography; limestone; dolomite; chalk; rock-typing;
porosity; permeability; rock-type; Abu Dhabi
Citation: GRANIER B. (2003).- A new approach in rock-typing, documented by a case study of layer-
cake reservoirs in field "A", offshore Abu Dhabi (U.A.E.).- arnets de Géologie / Notebooks on
Geology, Maintenon, Article 2003/04 (CG2003_A04_BG)
Résumé : Une nouvelle approche du 'rock-typing', illustrée par l'étude de réservoirs de type
'mille-feuille' du champ "A", domaine maritime d'Abou Dabi (Émirats Arabes Unis).- ns les
réservoirs carbonatés, la porosité Ø, résultant du cumul de différents types de volume de pores
(volume interparticulaire ou vacuoles non connectées, par exemple), et la perméabilité K sont deux
paramètres pétrophysiques faiblement corrélés. Aussi, parmi les différentes approches utilisées pour
estimer la perméabilité, celles qui combinent les paramètres pétrographiques avec les paramètres
pétrophysiques, connues sous le vocable de techniques de 'rock-typing', ont fourni les résultats les
plus probants. Toutefois dans de nombreux cas, de simples graphiques K/Ø sont encore utilisés pour
obtenir pour chaque 'rock-type' une fonction exponentielle ou logarithmique à partir de laquelle les
valeurs de K seront calculées en fonction de celles de Ø : ceci reste encore une règle quasi
générale, même lorsque le coefficient r2 s'écarte significativement de 1.
Le présent travail propose une approche basée sur une technique plus sophistiquée de classification
en groupes pétro-physico-graphiques, dans laquelle chaque famille est caractérisée par une
distribution 'log normale' de K. Cette technique facilite les changements d'échelle puisque elle
suggère d'utiliser pour chaque 'rock-type' la moyenne géométrique et le coefficient de corrélation
correspondant (variance ou écart-type) pour les valeurs de perméabilité.
Elle peut être appliquée avantageusement dans des puits non carottés en extrapolant la géologie
grâce aux logs électriques. A titre d'exemple, cette technique a été appliquée aux réservoirs de type
'mille-feuille' du champ "A" (Abou Dabi, Émirats Arabes Unis).
Mots-Clefs : Pétrophysique ; pétrographie sédimentaire ; calcaire ; dolomie ; craie ; rock-typing ;
porosité ; perméabilité ; rock-type ; Abou Dabi
1 ADMA-OPCO, P.O. Box 303, Abu Dhabi (United Arab Emirates);
present address: Total, DGEP/GSR/VdG, Tour Coupole, 2 Place de Coupole, La Défense 6 - Cédex 4,
92078 Paris La Défense (France)
Bruno.Granier@Total.com
Manuscript online since July 12, 2003
Introduction
In his pioneering work, G.E. ARCHIE (1950)
sets out the fundamentals of rock-type
classification. Any porous network is related to
its host rock fabric; therefore petrophysical
parameters, such as porosity (Ø), permeability
(K), and saturation (S), for any given "type of
rock" are controlled by pore sizes and their
distribution and interconnection. The goal of
reservoir characterization is to predict the
spatial distribution of such petrophysical
parameters on a field scale.
In field "A", offshore Abu Dhabi (United Arab
Emirates), the interval studied consists of a
series of layer-cake reservoirs bracketed by a
shale-dominated formation "A" above, which
forms the regional seal, and a mainly dense
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
2
limestone-dominated formation "F" below (LYON
et alii, 1998). In terms of reservoir, it has been
divided into 4 main zones (Fig. 1) denominated
from the top down "B" to "E". Sequence
stratigraphy helped delineate 16 "operational"
units (Fig. 1), of which the limits are sequence
boundaries, transgressive surfaces, or
maximum flooding surfaces. The layering used
in the latest reservoir model is based on a
combination of the overall stratigraphic
framework and the characteristics of the
reservoir rocks. It consists of ten layers from
top to bottom (Fig. 1):
• Layer 1, "B",
• Layer 2, "C",
• Layer 3, "C-tight" pars,
• Layer 4, "C-tight" pars,
• Layer 5, "D" pars,
• Layer 6, "D" pars,
• Layer 7, "D" pars,
• Layer 8, "D" pars and "E-tight",
• Layer 9, "E" pars,
• Layer 10, "E" pars.
Figure 1: Relationships of operational units and
rock-/litho- types to reservoir layers.
Although sequence stratigraphy helps to
establish a geologically valid reservoir
framework, rock-types are the building blocks
that complete the construction of reservoir
models. This study focuses on these building
blocks and more precisely on their petrophysical
parameters.
Figure 2: BRC/K cross-plot for the 65 samples used
for Hg injection. Caption: blue squares = discarded
samples; red dots = remaining samples.
Figure 3: BRC/K cross-plot for the 65 samples used
for Hg injection. Caption: blue squares = discarded
samples; red dots = remaining samples. The
program found a linear correlation between the two
parameters with r35 = 1.25 BRC + 0.07 with r2 =
0.86.
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
3
Figure 4: K/Ø cross-plot per layer (i.e. by "dominant" rock-type). All samples with a GEX (Gas Expansion)
porosity.
Figure 5: Hg injection data: "apparent pore-throat-size" (?m) versus "pore volume Hg saturated" (%). Key: red for
R-chalk ("B"); yellow for RR-0.25 ("C"); green for RR-0.45 ("D-E"); white for discarded samples. These should be
plotted as histograms; curves are used only for display!
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
4
Figure 6: Hg injection data: "apparent pore-throat-size" (?m) versus "pore volume Hg saturated" (%), cumulative.
Dashed horizontal line at 35% Hg injection. These should be histograms; curves are used only for display!
Figure 7: Hg injection data: "apparent pore-throat-size" (?m) versus "total volume Hg saturated" (%), cumulative.
These should be histograms; curves are used only for display!
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
5
Figure 8: Permeability histogram for RR-0.25 ("C") in layer 2 (all wells).
Figure 9: Permeability histograms for RR-0.25 ("C") in layer 2 (each well).
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
6
Figure 10: Permeability histogram for RR-0.45 ("D") in layer 7 (all wells).
Figure 11: Permeability histograms for RR-0.45 ("D") in layer 7 (each well).
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
7
Figure 12: Permeability histogram for RR-0.45 ("E") in layer 10 (all wells).
Figure 13: Permeability histograms for RR-0.45 ("E") in layer 10 (each well).
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
8
In his examination of the "rock type-
porosity-permeability relation", G.E. ARCHIE
(1950) stated that "a broad relationship exists
between porosity and permeability of a
formation" (a "formation" sensu G.E. ARCHIE is a
"type of rock" or "rock-type"). In his conclusions
he wrote, "The relations between rock
characteristics should be thought of as trends"
and added, "Actually, these may be expressed
by mathematical formulae". Neglecting the
other conclusions of this seminal paper later
workers drew trendlines on K/Ø cross-plots and
then used K/Ø transforms to predict K from
either the total or the effective Ø. The present
study addresses a different approach to the
"rock type-porosity-permeability relation".
Summary of the methodology
By definition, setting up a classification of
rock-type in subsurface reservoirs must be
based on cores. The approach is basically in 3
steps. First the several fabrics are identified
using thin section and conventional core
analyses. Then (or at the same time) the
preliminary types thus classified must be related
to Hg injection curves. In field "A", our database
includes:
• the rock-fabric (some 5000 thin sections
have been analyzed),
• the porosity range (some 2700 GEX /
Helium Ø and 2700 FLD Ø measurements
were available),
• the permeability range (some 3350 K
measurements were available), and
• the pore-throat size distribution derived
from the Hg injection curves (65 Pc curves
were available: Table 1, Fig. 3-7).
When such a procedure has been carried out
for cored intervals, it should be possible to
extrapolate it to other wells and to uncored
intervals using wireline information. In this
study, statistical analyses of available data were
undertaken before this last step took place
because they have helped greatly in making a
definitive rock-type classification.
Preliminary rock-typing (from thin
sections and conventional core
analyses)
First, it should be pointed out that in our
reservoirs there is little separate-vug porosity.
In any case, separate vugs do not contribute
significantly to permeability (LUCIA, 1983).
Second, some permeability measurements
must be discarded. Contractor plug descriptions
often mention the occurrence of joints (stylolite,
fracture, etc.). In such cases measured
permeabilities may be greater by a factor of ten
to a hundred than those obtained from samples
with similar fabrics but lacking these
phenomena. In this study, most K values higher
than 5 mD are the result of plug failure during
or before the measurement. Commonly they
represent less than 3 per cent of the samples.
Third, a plug is not necessarily homogenous;
it may include several rock-fabrics (e.g.
burrows, boundstone with matrix, …). Therefore
measurements made on a heterogeneous plug
should not be taken into account, even if the
thin section displays only one texture. For
example, plugs taken from the Rudist facies of
zones "C" to "E" commonly include large shell
fragments, so the matrix permeability (i.e. host
rock permeability) may be grossly undervalued
by conventional measurements ("Comet effect"
on K/Ø frequency maps per layer or on K/Ø
cross-plot: Fig. 4; "Manhattan effect" on 3D K/Ø
histograms: Fig. 14-15).
As a matter of fact, the range of
permeabilities encountered in this reservoir
study is very limited; consequently this small
breadth of field restricts the range in which the
several rock-types can be defined. Conventional
core studies and petrographical analyses
available for some 5000 samples have led to the
differentiation of 3 groups of lithotypes / rock-
types:
1. of very low permeability values (K < 0.5
mD2) and low porosity values (Ø < 10%),
2. of very low permeability values (K < 0.5
mD) and fair to high porosity values (Ø >
10%),
3. of low permeability values (0.5 mD < K < 5
mD).
Each of these three groups is comprised of 1
to 4 lithotypes.
The first group is made up of 4 lithotypes,
the poorest in terms of reservoir characteristics:
• R0-shale, the shale. It is characteristic
of the shale-dominated formation "A" above
the studied interval. It is also found mainly in
layer 3; there it forms the upper part of the
dense interval, formerly designated "C-
tight", which constitutes the effective barrier
between the so-called "B" and "C" reservoirs;
• R0-grain, the cemented grain-
dominated textures. It occurs mainly in layer
4; there it forms the lower part of the dense
interval, the former "C-tight", between the
so-called "B" and "C" reservoirs. The grain-
dominated textures
2 Darcy = 0.9869 µm2.
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
9
Figure 14: 3D K/Ø histogram for RR-0.45 ("D-E") (901 samples).
Figure 15: As above, K/Ø scales in reverse order.
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
10
•
are usually fully cemented (calcitic
cementation related to Echinoid
overgrowths). Locally some intergranular
porosity may be preserved in the basal fine-
grained grainstones but, due to pore-throat
cementation, this lithotype behaves as a
separate-vug system;
• R0-crypto, the mud- and wacke- stones
with a cryptocrystalline matrix. It is found
mainly in layer 8; there it forms a dense
interval, pro parte the former "E- tight",
between the so-called "D" and "E" reservoirs;
• R0-crypto + Separate Vugs, mud- and
wacke- stones with a cryptocrystalline matrix
and some very low to fair moldic porosity,
i.e. a separate-vug system. It occurs in the
predominantly dense limestone-dominated
formation "F" below the studied interval. The
common originally aragonitic bioclasts often
provided some moldic porosity;
unfortunately, the cryptocrystalline matrix is
not permeable.
The second group corresponds to R-chalk,
i.e. a nannofossil ooze matrix with fair to high
porosity, but very low permeability. It is
restricted to the zone "B", i.e. layer 1.
The third group includes 2 lithotypes, the
best in terms of reservoir characteristics:
• RR, the mud-dominated fabric (mud-,
wacke-, & pack- stones) with
microcrystalline matrix and fair to high
porosity values. The porosity is mainly
caused by the existence of a microcrystalline
matrix. It is the dominant rock-type in the
porous sections of zones "C" to "E", i.e. in
layers 2, 5, 7, 9, and 10;
• R-stylo, the mud-dominated fabric
(mud-, wacke-, & pack- stones) with
microcrystalline matrix and low to fair
porosity values. This lithotype differs from
the previous one because of heterogeneities
within the rock, i.e. bedding-related
stylolites that produce more tortuosity in the
porous network (and also possibly because of
a lesser vertical permeability range). It is
found mainly in layer 6; there it forms a
baffle within the so-called "D" zone, between
two reservoir sub-units (layers 5 and 7).
The classification by layers into permeable or
impermeable operational units has been set up
to take into account both sequence stratigraphy
(the "envelopes") and the dominant rock-type
(the "contents"). In order to simplify its
application (i.e. "up-scaling"), we chose a single
"dominant" rock-type to characterize each
layer. K/Ø cross-plots per layer (Fig. 4) and
petrographic analyses support this preliminary
rock-type classification. We note that RR
measures differ slightly between zone "C"
(above) and zones "D-E" (below); this is
possibly related to discrete depositional
environments and diagenetic overprints.
Towards a refined rock-typing (from
Hg injection data)
The permeability of a rock is controlled
primarily by the pore-throat network, i.e. size,
number, shape, and arrangement of pore-
throats:
• The larger the passages between pores,
the higher the permeability values.
Permeability is strongly weighted by the
larger pore-throats;
• For a given set of pore-throats, the
more numerous the passages between
pores, the higher the permeability values;
• Within a single rock-fabric / lithotype, it
is assumed that pore-throat shape is
homogenous and that pore-throats are
distributed randomly.
Pore-throat sizes can be estimated from the
Hg injection curves3 though these curves give
only an apparent distribution (WARDLAW &
TAYLOR, 1976).
Various cross-plots using factors4 derived
from the Hg injection curves (BRC, r35, etc.)
illustrate that pore-throat size is the main factor
controlling permeability (Fig. 2). The apparent
scatter of the data is partly due to the fact that
the several laboratories used divergent
increments in Hg injection pressure
measurements. These factors may also be
involved in the way the samples are sorted
(Table 1).
Here, 65 Hg injection curves were available
(but 4 permeability measurements were
missing because the corresponding plugs were
broken: Table 1); after screening, only 7
3 E.W. WASHBURN (1921) first suggested the use of
mercury injection as a laboratory method for
determining pore throat size distribution. Assuming
flow channels in the porous core sample may be
represented by a bundle of straight capillary tubes
in parallel with the spaces between the tubes
sealed by a cementing material, the WASHBURN
equation can be expressed as
Pc = -2 gamma
where Pc = capillary pressure (dynes/cm2), gamma =
surface tension of Hg = 480 dynes/cm, ? = contact
angle of Hg in air = 140°; and r = radius of aperture
for a cylindrical pore throat. Thus,
r (mm) =
4 r35 represents the apparent radius at 35% Hg
injection (PITTMANN, 1992) while BRC is an weighted
average radius also derived from the Hg injection
curve.
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
11
measurements were discarded (using as a rule
of thumb a BRC >> r35: Table 1, Fig. 3). When
considering the histograms "apparent pore-
throat size" versus "apparent percentage of
pore volume" per layer (Fig. 8-13), i.e. in their
relationship to the "dominant" rock-type, the
following points are significant:
• The first rock-type (group of lithotypes)
was all but unsampled as it characterizes the
tight zones;
• The second rock-type is poorly
represented, but 6 of 7 zone "B" samples (R-
chalk dominant) have a BRC < 0.15 ?m (r35 <
0.23 ?m);
• Most remaining samples (51)
correspond to the third group, possibly only
to lithotype RR;
• In this case, BRC values cover quite a
large range, from 0.15 up to 0.77 ?m (0.23
?m < r35 < 0.93 ?m), but it is possible to
discriminate 2 sub-groups within RR, one
with BRC < 0.3 ?m (RR-"C") and another with
BRC > 0.3 ?m (RR-"D-E").
Further investigations of rock-types
(using statistical analyses)
As mentioned above, there are variations
within lithotype RR that may justify splitting it in
accordance with the origin of the sample: zone
"C" or zones "D-E". However, thin sections in
carbonate rocks are usually not less than 30 ?m
thick; the largest pore entry radius based on the
Hg injection data used in the current study
never exceeds 2 ?m (and rarely 1 ?m).
Consequently pore throats were not visible in
thin sections. Therefore permeability and
porosity distributions were examined in each
layer (i.e. by "dominant" rock-type).
2D histograms (Fig. 8-13) show that:
• log K and Ø distributions follow
"Gaussian" laws within each layer of the
reservoir,
• average K for RR-"C" = 100.25 (= 1.78)
mD (Fig. 8-9),
• average K for RR-"D-E" = 100.45 (=
2.82) mD (Fig. 10-13).
Frequency maps and 3D histograms (Fig. 14-
15) show that:
• log K distributions are independent of Ø
distributions (grouped shotgun plot),
• trendlines observed on the K/Ø cross-
plot per layer / lithotype / rock-type are
misleading as they are highly weighted at
one end. Consequently the coefficient of
determination r2 departs significantly from 1:
o R-chalk ("B") K = 6.2 10-3 e0.2232Ø r2 =
0.4268 (438 samples)
o RR-"C" K = 25.6 10-3 e0.1958Ø r2 = 0.4596
(316 samples)
o RR-"D-E" K = 69.3 10-3 e0.1532Ø r2 =
0.5043 (901 samples)
where K (mD) and Ø (%).
In conclusion, RR has to be split into RR-0.45
(= RR-"D-E") and RR-0.25 (= RR-"C") which
form 2 distinct rock-types. A good match was
obtained with measured depth wireline log Sw
using a separate "Sw versus height" equation
for each rock-type:
• R-chalk ("B") log (Ø Sw) = - 2.159 Ø -
1.137 log h + 1.746
• RR-0.25 ("C") log (Ø Sw) = - 4.327 Ø -
1.175 log h + 1.367 log Ø + 3.727
• RR-0.45 ("D-E") log (Ø Sw) = - 0.729
Ø - 0.645 log h + 0.190 log Ø + 0.338
where h (ft), Sw and Ø (fractions).
R-stylo and RR-0.45 are considered a single
rock-type as the reduction in pore volume is
probably restricted to the immediate vicinity of
the solution seams. Presumably this reduction is
accompanied only by a correlative reduction of
the number of passages between pores (pore-
throats). Had it been accompanied by a
reduction in the size of the passages between
pores, the two lithotypes would have had to
remain distinct and therefore to be treated as
discrete rock-types.
Conclusion (towards a new approach
in rock-typing)
G.E. ARCHIE's classic paper (1950) was to
some extent misinterpreted. When he wrote:
"The relations between rock characteristics
should be thought of as trends. Actually, these
may be expressed by mathematical formulae",
he immediately added the following warnings:
"however, the formulae can not be applied in a
rigid manner (…). It must be kept in mind that
appreciable deviations from the average trend
may occur".
In many studies, rock-type classifications
end with trendlines drawn on K/Ø cross-plots
with K/Ø transforms used to predict K from
either the total or the effective Ø. However
statistics in this presentation show that the two
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
12
Table 1: List of the available Pc curves.
Carnets de Géologie / Notebooks on Geology - Article 2003/04 (CG2003_A04_BG)
13
ranges of data do not move together; the best
fits have low coefficients of determination (r2 <
0.5). This problem was attacked with a different
and promising approach: each rock-type
appears to be characterized by Gaussian
distributions of both log K and interparticle Ø,
which as determined here almost equals the
total Ø, and the two parameters are not
correlated (Fig. 14 -15). These distributions can
serve as a basis for rock-typing. Such an
approach has been used earlier to describe flow
in heterogeneous media (WARREN & PRICE,
1961). It is recommended that investigations be
continued regarding its feasibility when applied
to slightly more homogenous media, such as a
single rock-type, and how the use of a
geometric mean permeability value for a given
rock-type effectively affects reservoir models
and simulations.
Acknowledgements
The author thanks the Management of Abu
Dhabi Marine Areas - Operation Company
(ADMA-OPCO) and Abu Dhabi National Oil
Company (ADNOC) for their permission (Ref.
No.: E/OFFSH/SPG/530/99) granted to publish
this paper on the occasion of the 9th ADIPEC
(2000). Due to a large number of speakers
attending the conferences the organizing
committee decided to retain only one of the two
papers offered by the author (the selected
paper concerns the stratigraphy of the
Thamama: GRANIER, 2000). Subsequently the
remaining paper was submitted to a foreign
geoscience paper-printed journal and was
accepted for publication. Two years later the
journal was still not issued. Therefore it is to
appear in an electronic form in "Carnets de
Géologie - Notebooks on Geology". Reviews of
the preliminary versions by John BELLAMY and
Trevor BURCHETTE resulted in many
improvements to the paper. Special mention
goes to Nestor J. SANDER.
References
ARCHIE G.E. (1950).- Introduction to
petrophysics of reservoir rocks. Bulletin of
the American Association of Petroleum
Geologists, Tulsa, vol. 34, N° 5, p. 943-961.
GRANIER B. (2000).- Lower Cretaceous
stratigraphy of Abu Dhabi and the United
Arab Emirates - A reappraisal.- 9th Abu
Dhabi International Petroleum Exhibition &
Conference, Conference Proceedings,
ADIPEC 0918, Abu Dhabi, October 15th-18th,
p. 526-535.
LUCIA F.J. (1983).- Petrophysical parameters
estimated from visual descriptions of
carbonate rocks: a field classification of
carbonate pore space. Journal of Petroleum
Technology, Houston, vol. 35, Nº 3, p. 626-
637.
LYON T.A., FULLER J., GRANIER B., HOZAYEN M., AL
RIYAMI A., KHALAF A. & THIÉBOT B. (1998).-
Integrated study of a faulted and fractured
reservoir.- 8th Abu Dhabi International
Petroleum Exhibition & Conference,
Conference Proceedings, SPE 49453, Abu
Dhabi, October 11th -14th, p. 45-67.
PITTMANN E.D. (1992).- Relationship of porosity
and permeability to various parameters
derived from mercury injection-capillary
pressure curves for sandstone. Bulletin of the
American Association of Petroleum
Geologists, Tulsa, vol. 76, N° 2, p. 191-198.
WARREN J.E. & PRICE H.S. (1961).- Flow in
heterogeneous porous media. Society of
Petroleum Engineers Journal, Houston,
September 1961, p. 153-169.
WARDLAW N.C. & TAYLOR R.P. (1976).- Mercury
capillary pressure curves and the
interpretation of pore structure and capillary
behaviour in reservoir rocks. Bulletin of
Canadian Petroleum Geology, Calgary, vol.
24, N° 2, p. 225-262
WASHBURN E.W. (1921).- Note on a method of
determining the distribution of pore sizes in a
porous material. Proceedings of the National
Academy of Science of the United States of
America, Washington, vol. 7, p. 115-116.