Professional Documents
Culture Documents
hellénique
Dinsmoor William Bell. Studies of the Delphian treasuries. II : The four Ionic treasuries.. In: Bulletin de correspondance
hellénique. Volume 37, 1913. pp. 5-83.
doi : 10.3406/bch.1913.3131
http://www.persee.fr/doc/bch_0007-4217_1913_num_37_1_3131
.
heavier material, we find a development which is unknown
in the poros structures or the limestone foundations. The
lead dove-tail clamp is now reenforced by a bar of iron,
bent down at both ends and forming a hook clamp which
fits into vertical borings in each end of the dove-tail
cutting (figure 1) (I). It would appear therefore that at
(\) Such clamps have been found in many sites of Asia Minor, but
have been almost always misinterpreted, as a dove tail clamp of iron
pierced at each end for an iron dowel (Henderson, British Museum
Excavations at Ephesus, 280; Wilberg, Œst. Forschungen in Ephe-
sos, 224-225; Koldewey, Insel Lesbos, 46; Durm, Baukunst", 146, 164),
a form which, so far as I know, never existed. The hook clamp in
the dove-tail cutting traces its origin back to Mycenaean times; it is
found in the «Treasury of Atreus» at Mvcenae (Perrot and Chipiez,
VI, 615, 627, 628; Koldewey, Insel Lesbos, 46); it afterwards became
peculiarly Ionian, and appeared in the temple of Apollo at Nape on
Lesbos (Koldewey, Insel Lesbos, pi. 16, 9-15), the Croesus Temple at
Ephesus (Wilberg, Œst Forsch.-in Ephesos, 228, fig. 192; British
Museum Excavations at Ephesus, atlas, pi. X, text, 259, 280), in the
temple on the island called Palati at Naxos, in the Acropolis temple at
Paros, in the temple at Assos (Clarke, Investigations at Assos 1882/
1883, 61, 66), in the Doric treasury IV at Olympia (Olympia, I, pi.' 34),
and in the palace of Darius at Persepolis (Perrot and Chipiez, V, 470,
fig. 298; Dieuiafoy, L'Art antique de la Perse, I, fig. 16) — though it
must be noted that even in this region the form had appeared much
earlier, at Khorsabad (Place, Ninive et V Assyrie, pi. 70). At Delphi
itself they appear not only in the four Ionic treasuries, but also in
10 STUDIES OF THE DELPHIAN TREASURIES
the beginning iron was not the main part of a clamp,
but the reenforcement, especially when used in marble
construction ; and lead, far from being the mere sealing
for the iron clamp, as in later times, was first used alone,
the direct successor of wooden dove-tail clamps such as
we find in Egypt. The iron was always omitted in po-
ros construction, and sometimes even in marble, in work
at Delphi down to the end of the sixth century; thus
in some blocks of the Siphnian entablature only one of
the two clamps at each joint has the iron reenforcement.
Though the general form is always the same, these clamps
vary in dimensions ; the Cnidians used a small size, about
0.12 m. long, 0.05 m. wide at the ends and 0.02 m. at
the waist, very shallow and ill defined; for Clazomenae
and Massilia they are slightly larger, 0.15 m. long, and
more sharply cut; in the Siphnian Treasury we find the
largest of all, 0.20 m. long, 0.04 m. wide at the waist
and 0.08 m. at the ends.
Dowels. — Dowels on the other hand are rare in all
except the Cnidian Treasury. But in this the bottom of
every block is do welled at one end; the cuttings are of Τ
form, about 1 cm. deep and evidently intended for molten
lead only, because they are not outlined sharply enough
to receive iron dowels; the lead was poured in through
the bottom of the T, which reached to a joint, lisualy δ
or 6 cm. but sometimes 15 cm. from the head of the Τ
(figure 1). Dowels were used in the Cnidian and
Siphnian Caryatides and in the columns of the Massiliot
Treasury (figure 3). Both of the marble Doric treasuries have
every block dowelled.
the two earliest marble Doric treasuries, that of the Athenians and
the κάτω ναός at Marmaria. Later examples are found in the Ionic
temple at Messa (Koldewey, Insel Lesbos,, pi. 22), and in the Doric
temple of Athena Polias at Pergamon (Alterth. v. Pergamon, II, 21,
pi. 8); in these, as in the examples at Olympia and Assos, the true
character as a hook clamp with a lead dove-tailed covering was
recognized (Koldewey, Insel Lesbos, 46).
STUDIES OF THE DELPHIAN TREASURIES 11
Wall Base, Torus. — On the foundation at Marmaria,
the only one of the four buildings which is sufficiently
preserved to show this feature, rests a socle composed
of two courses of marble, a plain plinth and a moulded
torus, which form the decorative base of the building.
The moulded torus, on the façade, becomes the stylobate
at Marmaria ; this was the case also in the Cnidian
Treasury, but in the Siphnian the torus rested on the stylo-
bate, which was then the plinth course below the torus·
In no case is there any arrangement of steps or other
means of getting access to the treasuries ; yet in that of
Massilia, the stylobate is 0.855 m. above the ground
level (1), and from what little evidence we have, it is
certain that the treasury of Siphnos was similarly arranged
(figure 3). Indeed, this seems to have been the customary
scheme at Delphi (2) except in the case of the poros
treasuries, of which the stylobates were raised only one step
above the ground; the marble Athenian Treasury has a
stereobate of three courses with impossible treads of
only 0.059 m. and 0.048 m., raising the stylobate 0.903
m. above the paved platform before it; the limestone
Theban Treasury had likewise three receding courses,
with treads of 0.065 m. and 0.155 m. bringing the
stylobate 0.80 m. above the euthynteria; and in the later
Syracusan Treasury the lowest «step» projected only
about 0.10 m. beyond the stylobate.
(1) Identified by its technique, and also because on the façade the
torus returns for only 0.495 m., agreeing perfectly with the width of
the parastades which are certainly, on epigraphieal grounds, to be
identified as Cnidian. -
(2) One built into the ruined chapel of St.jGeorge below the precinct,
a second just outside the east precinct wall, near treasury XIII, the
third on treasury XIII itself.
STUDIES OF THE DELPHIAN TREASURIES 13
assign a type which comes second chronologically, and
cannot be assigned to any of the three other buildings;
it is moreover very rare, for I have seen only the small
piece in the Museum. The block is 0.239 m. high,
decorated with a channeled torus combined with a beaded
astragal; the torus, 0.184 m. high, is not semicircular
but parabolic in section, with greater fullness toward the
bottom, projecting only 0.065 m. ; the upper two thirds
contain four channels, while the lowest third has a plain
surface. This base moulding was used by Mr. Tournaire
in his restoration of the «Cnidian» (Siphnian) Treasury
(Fouilles, II, pi. 11). At Marmaria we find a similar base
in situ, in the treasury which I assign to Massilia ; the
torus here, 0.215 high, is decorated with eight channels,
all of which are hollowed ; the beaded astragal is in this
case cut on the lowest wall block. The torus forms on
the façade the stylobate, as in the Cnidian Treasury ;
and in both, strangely enough, the exposed and now
footworn stylobate shows unconcealed clamps. Finally, a
torus which is itself cut in the form of colossal beads
and reels, and therefore needed no beaded astragal above,
must be assigned to the treasury of Siphnos. At first
unidentified (BOH, 1896, 589; 1897, 303), it was finally
recognized as the base moulding of the « Cnidian »
(Siphnian) Treasury (BCH, 1900, 603, and the model at
Delphi, figure 12); but in this it has been placed in the
stylobate course (1), whereas it is actually the successor of
the small beaded astragal in the Massiliot Treasury, and
should be placed above the stylobate so as to be
carried around the interior of the prodomos. This is shown
by two facts; first, the threshold of the Siphnian
doorway is decorated with the same colossal beads, which
(1) I may note here that the details (figures 4 and 11, are
reproduced at Vs full s'ze; Plate I is at half this scale, or ljm full size; and
for purposes of comparison figures 3, 5-9 are all reproduced at lf32
full size; fig. 10 is at no scale.
STUDIES OF THE DELPHIAN TREASURIES 23
(1) Mr. Bourguet has samples of each (Fouilles, III, i. 155); he gives
the heights as 0.465 m., 0.485 m., 0.50 m., and 0.52 m.
(2) Mr. Bourguet recognizes only one of these, 0.204 m. high.
(3) In view of this it must be said that Dr. Pomtow's statement
(BPW, 1909, 190) that the walls of the Athenian Treasury have been
reconstructed in great part with wall blocks of the Cnidian is utterly
without foundation (cf. Bourguet, BGH, 1910, 225); his reason for
making the statement was, as he himself agrees, his inability to find
the true Cnidian blocks; cf. BPW, 1909, 189-190, where he assumes that
they are of the same height as the courses of the Athenian Treasury ^
average 0.376 m.
26 STUDIES OP THE DELPHIAN TREASURIES
:- H
Κ --
Epistyle
Stjrlobate-
v 9 c /v
^JiMfe
(fig. 6). The lower inscription begins only 0.14 m." from "the joint
surface; according to Bourguet's restoration of line 4 of no. 309, with
105 letters (17 letters occupy 0.175 m.), the inscription would be only
1.08 m. long, and the joint surface only 1.22 m. from the end of the
BULL. DE COERESP. HELLÉNIQUE, XXXVII 3
34 STUDIES OP THE DELPHIAN TREASURIES
parastas, not agreeing with the upper inscription. But Fouilles, no.
309 may be restored as follows:
(1) With one variation, the 0.475 m. and 0.494 m. courses being
interchanged. This system of a regular decrease is followed by
Bourguet and Martinaud (Fouilles, III, i.), interchanging however the
courses of 0.544 m. and 0.504 m. because of the Cassander Inscription.
Pomtow suggests a regular decrease of 0.022 m. in each course from
0.59 m. to 0.37 m. (BPW, 1909, 188), which would require at least
eleven plain courses.
STUDIES OF THE DELPHIAN TREASURIES 41
wo WA*r3$&&*%£%4
2 /Λ.
JIorth
Figure 7. North-east Corner of the Siphnian Treasury.
42 STUDIES OP THE DELPHIAN TREASURIES
(1) I do not however believe that one of these blocks can be actually
placed directly on the other; the lower block should be farther to the
right.
(2) At Priene and Pergamon the plinth courses of the wall are
separated by two courses of orthostates, but in these cases the true
construction is suggested by the heights of the blocks.
STUDIES OP THE DELPHIAN TREASURIES 43
0.371 m. high, seem to have aligned with the eighth wall
course (average 0.360 m); the others are all smaller,
either 0.255-0.261 m. or 0.305-0.306 m., and cannot have
aligned with any course of the exterior. But on figure
5, it appears that the third, fourth, and fifth courses,
all orthostates, together occupy a height of 1.513 m.,
which corresponds to six courses of 0.252 m.,
approximately what we obtain as one type of the antithemata.
Acting on this principle of two backing courses to each
of the orthostates, the blocks 0.305 to 0.306 m. in height
might, with others slightly lower, make up the total
required for the first course, 0.592 m. Like the plinths,
these orthostates have clamp cuttings for the revetment
which concealed this irregular construction.
As in the case of the Cnidian Treasury, I have
compiled a list of the extant Siphnian wall blocks to
supplement the publication by Mr. Bourguet in the Fouilles
de Delphes, III, i. 113-149; compare the plate with the
partial assemblage of the stones accompagnying his text
Siphnian Treasury.
4- H
Ί λ .'· ,ΚΙΙΊΪΙ
'
%/''%;λ./^,Α,/^../,Α
,
according to the length of the verb, we should require
in addition to the letters preserved:
32 letters (άνενεώσαντο, with τάν) ;
29 letters (άνενεώσαντο, omitting τάν ; or άπέδωκαν with
τάν) ;
27 letters (έδωκαν with tctv) ;
26 letters (άπέδωκαν without τάν);
24 letters (έδωκαν without tctv).
We must choose between the five lengths made possible
on epigraphical grounds, 2.565 m., 2.415 m., 2.315 m.,
2.265 m., or 2.165 m. (1). The spacing of the ornament,
making the inscribed fascia 2.557 m., 2.333 m., or 2.110
m. long according as we restore eleven, ten, or nine
repeats of the lotus-and-palmette ornament, will coincide
with the indications given by the inscription only when
the reading is άνενεώσαντο tctv, or, less perfectly,
έδωκαν tctv. And only with the former can we
obtain reasonable space in the second line for the names
of the bouleutai (Plate I). Of Mr. Homolle's readings I
therefore prefer
■
'
Various Arrar?Qen?ei?l5
of Frieze &-Epi*'yIe
(1) Cf. BGH, 1897, 303; Courby goes at length into the question of
this moulding, RA, 1911, 208-209.
(2) Both were found in the mass of Cnidian fragments east of the
Roman court before the entrance to the precinct. The angle piece now
lies west of the Potidaean Treasury.
STUDIES OP THE DELPHIAN TREASURIES 65
(1) While giving the sculptured frieze in the main to the Siphnians,
Pomtow and Bulle have recently (BPW, 1911, 1613) ascribed to the
Cnidians some fragments which are really from the Siphnian and
Massiliot friezes.
BULL. DE CORRESP. HELLÉNIQUE, XXXVII 5
G6 STUDIES OP THE DELPHIAN TREASURIES
sess twelve large pieces and a great number of
fragments. The problem has long been whether we should
call the frieze Siphnian or Cnidian. Recently the matter
has been somewhat complicated by Prof. Heberdey's
attempt to divide the frieze and give a portion to each
of these two treasuries (AM, 1909, 145-1GG). But any
method of breaking up a frieze which is so obviously a
unit, with a uniform height 0.G40 m. (except two slabs
which increase from 0.G40 m. to 0.G80 m. with a
regular curvature), uniform in all its other dimensions and
in its technique, presents at once insurmountable
difficulties which Prof. Heberdey has not overcome. The
differences are only such as might naturally occur in the
work of two separate sculptors working on the same
composition. It can be conclusively shown that the wiiole
belongs to one building (cf. Part III) (I). As for the older
question, whether it is Cnidian or Siphnian, wo have
already found a blank Cnidian frieze which supplants
the claims of the sculptured slabs; the latter show the
treatment of beds and joints seen in the Siphnian
Treasury, and on them appear the marks of the toothed chisel,
which is absolutely unknown in Cnidian work. The tops
of the frieze slabs fit, piece by piece, the blocks of tho
Siphnian cornice; the bottoms showed traces of the eggs
(BCH, 1897,- 303) which crowned the Siphnian epistyle.
The Siphnian frieze is treated in a manner analogous
to the earliest sculptured Doric metopes, such as the first
and second series at Selinus. The entire scene is enclosed
within a rectangular frame which projects practically to
the outermost plane of the sculpture; only along the top
is it omitted and its place taken by the frieze crown.
The background is not, as ordinarily, the main line; here
it is the frieze frame, strongly marked at the corners of
the building, and the sculpture is sunk within it. Such
(1) Such is the result also of the recent studies by F. Courby (RA,
1911, 197-220), and Pomtow (BPW, 1911, 1G12).
STUDIES OF THE DELPHIAN TREASURIES 67
(1) Courby (RA, 1911, 209) defends the Museum restoration and
even increases the projection of the plinth beyond the egg and-dart,
from 0.03 m. to 0.042 m., because of an incised «setting line» on the
bottom of a slab of the Gigantomachy. I may note that on the slab
with the altar (Fouilles, IV, pi. 9-10, 1) there is a «setting line» as much
as 0.057 m. back from the edge of the plinth. I do not believe that
either, however, is connected with the process of setting the frieze
blocks on the then uncut egg moulding. The interesting detail of the
hole pierced for water in the plinth (RA, 1911, 209, fig. 2) does not
seem to require such a projection; it would better deliver the water
at a point hidden in the depths of the carved egg-and-dart.
(2) The angle treatment of this moulding was more complex than
that restored in the model, as is shown by the top of a corner now
in the magazine of the Museum.
(3) In the magazine of the Museum is a fragment of a leaf-and-dart,
similar in style to that of the Siphnian Treasury, but somewhat
smaller (0.126 m. in spacing) though the height of the beaded astragal,
STUDIES OP THE DELPHIAN TREASURIES G9
At Marmaria, as noted by Homolle and Poulsen, were
found many pieces of leaf-and-dart which evidently
ts
Cr)ldu5 ψ,
the lower bed surface is wider than usual, 0.035 m., while
on the top it is only 0.025 m. wide; therefore whatever
came above was either flush with the face of the abacus,
which seems improbable, or projected beyond it. The
latter alternative is favored by the position of the τ
dowels, which are very close to the face of the abacus
(centered about 0.065 m. distant). This must be therefore
the bed moulding of the geison ; a similar profile was
used in the same position in the Croesus temple at Ephe-
sus (British Museum Excavations at Ephesus, atlas, pi.
10, 15). The spacing of the eggs varies from 0.1235 to
0.127 m., exactly twice that of the bead on the top of
the Cnidian frieze.
Geison. — Dr. Pomtow proposes that, because in the
ornament of the Siphnian door enframement the palmettos
are broad, the type with the narrow palmettes, appearing
on certain fragments of geisa, should be identified as
Cnidian (BGH, 1909, 188). The geisa to which he refers,
however, come from the Massiliot Treasury. I have seen
no fragments which could be attributed to Cnidus.
At Marmaria, and even in the precinct of Apollo, were
found many fragments of richly decorated geisa, with
the alternating lotus-and-palmette in the soffit, belonging
to the Massiliot Treasury. The beaded astragal is set up
into the undercut soffit, so that the lowest point of its
circumference is flush with the bed of the geison. We
find three types all with, the above characteristics and
with the same projection 0.204 m. from the edge of the bed:
1) the lotus has a calyx, the upper surface slopes, and
the height of the face is 0.157 m. ; this was the geison
from the flanks.
2) the between
height' lotus as beds
before,
andthe
of upper
the faceside
0.157
horizontal,
m. ; this was
the
to the lower bed, the height of the face is 0.143 nr; this
was the raking geison of the pediment, as identified by
a fragment of the apex in the magazine of the Museum.
The spacing of the repeats of the ornament is 0.180 m.,
so that the spacing of the beaded astragal is 0.060 m.,
exactly the same as that of the beaded astragal which
crowned the Massiliot frieze.
Almost exatcly like the geison crowning the door of
the Siphnian Treasury are the numerous fragments of
what have always been regarded as the main geison of
that building (shown in Perrot and Chipiez, VII, 649 ;
Journal RIB A, 1(JO4, 33; Durm, Baukunst*, 332). The
larger pieces were all found near the Siphnian
foundation; this fact, the technique, and the exact reproduction
of the door ornament, all indicate that they are Siphnian.
The rosettes which appear on the face of the door geison
are here omitted. In the lotus-and-palmette ornament of
the soffit the lotus is always sheathed, and the beaded
astragal projects below the bed for a full radius. There
are three varieties;
1) that from the flanks of the building, with sloping top
and cuttings for rafters and eaves simas, the face 0.173
m. high, the projection 0.281 m.
2) that which formed the pediment floor, with horizontal
top and no trace of the roofing, 0.128 to 0.134 m. high
between beds, the face 0.173 m., the projection 0.281 m.
3) the raking geison of the pediment, of exactly the same
type, but 0.122 to 0.128 m. high between beds, the face
0.144 m., the projection 0.301 m.; identified by a bevelled
piece for a lower angle of the pediment, in the Museum.
Tympanum. — On the tops of the geisa which formed
the floor of the pediment is a slight rise of the surfaco
0.06 m. from the face; outside, this surface is much
weathered, while inside the rise it formed a bed for the
tympanum and was entirely protected. Now the endmost
blocks of the pediment floor, the angle geisa, have the
STUDIES OF THE DELPHIAN TREASURIES 73
extreme end of the raking geisa cut on the same block to
avoid a feather edge; these pieces show that the raking
geisa did not extend forward flush with the horizontal
geisa, as in later times, but were set back 0.024 m. (the
same is true, for instance, of the Athenian Treasury, where
the setback is 0.013 m.). From the front of the soffit of
the raking geison back to the beginning of the bed which
rested on the top of the tympanum is 0.295 m. The top
of the tympanum was therefore 0.295+0.024 = 0.319 m.
inside the edge of the pediment floor; the foot of the
tympanum was, according to the weathering, only 0.060 m.
inside, seemingly a discrepancy of 0.259 m. But it so
happens that in the only completely preserved tympanum
which can be assigned to this series of treasuries (Fouilles,
IV, pi. 16-17, 1) a plinth at the bottom, on which the figures
stand, extends forward exactly 0.260 m. from the top
of the tympanum (figure 10). A better fit would be
impossible ; but it is more than a question of mere fitting. In
spite of the archaic style of the sculpture, the treatment
has absolutely nothing Cnidian about it ; the background
is tooth-chiselled and, while I could not examine the lower
bed, a fragment of the right end block of the other
tympanum, now in the depot north-east of the Museum, shows
the Siphnian treatment with the broad smooth bearing
surface at the edge. It therefore appears that this archaic
tympanum must belong, to the same building as does the
Siphnian frieze, and that in this respect the model is
correctly restored. To the other tympanum of the Siphnian
Treasury may be assigned some of the fragments of
sculpture in the Museum, and the piece of an angle block
northeast of the Museum.^*
The complete Siphnian tympanum is composed of three
pieces, a central block (κορυφαίος) 1.641 m. long, and on
either side a κερκιδιαΐος, at present 1.60-1.64 m. long, the
extreme ends having been broken off; the total length is
then 4.88 m. But from the height measured at the center,
74 STUDIES OF THE DELPHIAN TREASURIES
(1) Central block 1.641 in. long; height at center 0.735 m., at left end
0.530 m. and at right end 0.524 m.; average height at ends 0.527 m.,
slope 0.735-527 = 0.208 m. in 1/2x1.641 =0.8205 m.
(2) Calculated by Mr. Homolle as 5.78 m. (BCH, 1896, 588), and this
result used by Prof. Heberdey (AM, 1909, 147).
(3) Prof. Hebérdey made this geison height, in his calculation, 0.17 m.;
he evidently measured the nosing of the horizontal geison, 0.173 m.
high. The width of the treasury on the horizontal geison then became
5.78+(2x0.66) = 7.10 m., too much for the foundation (AM, 1909, 147).
STUDIES OF THE DELPHIAN TREASURIES ' 75
W. B. DINSMOOR
ADDITIONAL NOTE
After the printing of Parts I and II, Dr. Pom tow's article on clamp
forms in Delphi (BPW, 1912, 636-640) came to hand, and seems to
call for some comment. The dove-tail clamps were never of iron, as
Fiechter suggests, nor of wood, as Pomtow says was the case when
the circular borings («Dornen»; cf. Part II, fig. 1) were omitted;
dovetail clamps of the Greek period in Delphi were always of lead, as is
attested even by. the Siphnian foundation which Pomtow cites as the
chief example of clamping with wood; iron was used only when the
bored holes were present, and then in the form of η hook clamps (cf.
p. 9 above). In the table of buildings with dovetail clamps (I. c. 637),
it appears that this form has now been correctly recognized in the
«Sicyonian prostylos» or early Syracusan Treasury (ei.BGH, 1912, 468
(1) f or purposes of comparison, figures 12 and 13 are reproduced
at approximately the same scale, 1/160 actual size.
STUDIES OF THE DELPHIAN TREASURIES 83
n. 1). In the table of buildings with h clamps (I. c. 639-640), the Cni-
dian Treasury is named because Dr. Pomtow would place it on
foundation VII, disregarding the fact that the h clamps occur only in the
added south wall or in repaired blocks along with far from
contemporary dove-tail clamps. Finally, I think that an investigation of the
«repair-clamps» on the top of the Cnidian dedicatory inscription,
included in his list of π hook clamps (I. c. 638-639;, will reveal the fact
that the cuttings were made later than 1896 for the purpose of
exhibiting the fragments.
W. B. D.
May 24, 1912.