² Greek Buildings Represented in the British Museum, 148. ‘ I do not think it has been noticed that the temple of Rhamnous …is so like the Theseum (viz. Hephaisteion) that one must be a copy of the other.’

³ For this, see, e.g., Dinsmoor, Hesperia, Suppl. V, 153.

⁴ I use this term of the space between the outer colonnade and the wall of the cella buildings. In doing so I follow Dinsmoor (Hesperia, Suppl. V, 59) and Stevens, (AJA XV, 18Google Scholar). Durm, (Baukunst, 1892, 146Google Scholar, et al.) refers to the same part of the building as a ‘pteron’. There seems to be no ancient support for either usage, as Ebert insists in the case of ‘pteron’. See his Würzburg, dissertation, Der Tempel (1910), 6.Google Scholar

⁵ AA 1929, 447 ff.

⁶ Thompson has recently proposed a complete restoration of the East Pediment of the Hephaisteion: see his article in Hesperia, 1949. My drawing, which I made just before his discovery of the figure of Herakles (Spring, 1948), will not permit me to show any of his restoration. I hope one day, if it wins general acceptance, to include it in a more complete study of the temple than that I am at present offering. For the pedimental sculptures of Sounion, see, e.g., Stais, in AE 1917, 198 and Fig. 11.Google Scholar At Rhamnous, on the other hand, Orlandos believes the pediments were empty. (BCH 1924, 317.)

⁷ Principles of Athenian Architecture, pl. XXXV and Chapter XI. Though he shows little, he shows it most accurately.

⁸ ‘The Ceiling of the Opisthodomos of the Theseum’, AJA XV (1911), 18–23.

⁹ This is apparently Lethaby's ‘joggle-jointing’ perhaps shared with the Hephaisteion by fourth century Ephesus (Lethaby, op. cit., 21). I find no examples figured in Durm, and was puzzled by these joints until enlightened by Mr. Hill.

¹⁰ viz. pronaos, cella, and opisthodomos combined.

¹¹ By this term I mean half the normal interaxial distance between the outer columns. The unit determines the length of the stylobate blocks, of nearly all blocks in the entablature and usually of the roof tiles as well. The Hephaisteion (see below, p. 75) is a rare exception.

Tilton lap. Waldstein, , Argive Heraeum I, 120Google Scholar) refers to one Greek foot as a ‘unit’. I wish to make it clear that I am not using the word in Tilton's sense.

¹² Taking, for instance, the twenty-four marble blocks of the second step on the north side and excluding the east and west corner blocks (2–07 and ca. 2·10 respectively), one finds a line of twenty-one between 1·28 and 1·31, and only the twenty-second, at the extreme west, as short as 1·265. This seems to me to show the course was laid from east to west. It would be rational to lay out the important east end first, with the greatest possible precision of jointing (see also below, p. 70).

¹³ Another term that worries me. I use it here, following Dinsmoor (33 ff.) and Robertson (Greek and Roman Architecture ², 41), of the course below the lowest step. I would prefer to use it of the horizontal plane forming the lower boundary of the visible building : a plane formed sometimes by levelling rock, sometimes by building up foundation courses. I understood Mr. Hill on the Acropolis to use it in this sense. But the low marble platform at Sounion, which Dinsmoor and Robertson would describe as a euthynteria, forms by contrast an integral part of the visible building. In ancient usage, as shown by IG II² 1668, 16 ff. (the specifications for Philo's Arsenal), means what we call a toichobate:

¹⁴ ‘The purpose (of a poros lower step) obviously was, as previously in the Older Parthenon and later at Rhamnous, to effect a transition from the marble to the surrounding terrace.’

¹⁵ The Kara stone of the Older Parthenon was distinctive and beautifully worked, and formed a true transition from marble to poros. See Hill, AJA 1912, 535 ff., together with his figs. 3, 12, and 13.

¹⁶ III, 3, 11.

¹⁷ Expédition Scientifique de Morée (1831) III, pl. 32.

¹⁸ This figure, indeed, is entailed by Dinsmoor's other dimensions, and I reached it from them for myself by taking the figure 0·0475, the distance at which, accord ing to him, the outermost arrises of the column stood back from the face of the stylobate (IX, 24).

²⁰ For Dinsmoor's increase of the corner column, see above, p. 70.

²¹ Op. cit., 43.

²² Perrot and Chipiez VII, fig. 233, III. As Greek Doric architecture developed, it is on the whole true to say that the thickness of the epistyle, originally less than the lower diameter of the columns, came by degrees to equal and then to exceed it.

²³ Penrose, Principles of Athenian Architecture, pl. XXXV; cf. Dinsmoor IX, 31.

²⁴ Dinsmoor, in AJA XLIII, 32. But I am not quite sure of this figure, as Dinsmoor may be measuring not from the corona but the rim of the ‘pedimental step’.

²⁵ See below, p. 86.

²⁶ For a slight difficulty in the evidence for Sounion, which does not, however, affect the main lines of the structure, see below, p. 86 and note 63.

²⁷ Dinsmoor, 84. I use the term ‘plinth course’ of any wall-course between the orthostates and epikranitis, thus departing from Robertson, who uses it of the course sometimes found between the toichobate and orthostates, and called by Dinsmoor (42) ‘the moulded wall-base’. Robertson agrees that the courses above the orthostates were originally of mud-bricks or ‘plinths’ as in the Olympian Heraion (Robertson², 62), and I think it best to confine the term ‘plinth’ in Greek Architecture to these courses.

²⁸ Thé system is best understood by studying Cock-erell, Aegina and Bassae, especially the beautiful drawing on Bassae, pl. VII. See also The Unedited Antiquities of Attica, ch. VI, pl. XII.

²⁹ The central joint of the flank sima was in line with that of the geison (Dinsmoor, 114). Measuring from this to one corner, one obtains a total of 12 × 1·299 + 0·67, or 16·26. This, with a discrepancy of less than one centimetre, is the length of half the flanking geison.

³⁰ The fragment appeared too late (cf. Dinsmoor, 113, n. 258) for Miss Shoe's exhaustive study, Profiles of Greek Mouldings (1936).

³¹ See below.

³² For this term, see above, n. 4.

³³ He uses it of those at Sounion in AE 1917, 223, fig. 18, while in BCH 1924, 316–7, he uses the term of the whole epikranitis, both of the cella and the flanking colonnades. This is reasonable.

³⁴ Robertson's ‘plinth-course’. See above (n. 27).

³⁵ This moulding was not continued internally even in the opisthodomos, where it would have been visible. But like Stuart and Revett I have carried it round the inner faces of the antae. At present it stops on either side about two-fifths of the way along them (Dinsmoor, figs. 19 and 34). But the profile of their west faces (fig. 19) seems to show it once extended as far. It was perhaps interrupted by one of those mysterious ‘sills’ certainly present at Sounion (plate 8 to this paper) and restored by Dinsmoor (fig. 34) in the Hephaisteion.

³⁶ See above, n. 27.

³⁷ I have calculated this from Dinsmoor, 41. He is giving lengths for the toichobate block: his figures for the east end are simple. 1·186 minus 0·097 gives the length of the anta, 1·09. Emending his figure of 1·75 for the west end to 1·175, one finds that 1·175 minus 0·08 gives an identical length, 1·09, for the west anta.

³⁸ See below, p. 92.

³⁹ See Penrose, pl. XVI.

⁴⁰ The arrangement at Sounion was the opposite. There the door projected, if at all, towards the cella. See below, p. 93.

⁴¹ I have nothing to add to the evidence of Stevens' scratch (Dinsmoor, 9 1 ), which seems to me on the whole persuasive. With the presence of a west colonnade, one can understand the excess of the most westerly space (1·47) over the most easterly (1·15).

⁴² ‘The so-called Theseon’ (sic) …‘ supremely uninteresting in itself … Why should this temple, the best preserved of its date in the world, built within a few years of the Parthenon, and embodying all the same principles, remain by comparison so devastatingly boring?’ (Classical Landscape, 47.)

⁴³ The correct dedication is due to Dörpfeld's identification of the temple of Athena; see AE 1900, 122 ff.

⁴⁴ ‘Indeed, from the motor road it appears, when silhouetted against the sun, remarkably like a half-finished hangar.’ (Op. cit., 87.)

⁴⁵ Refinements, however, perhaps existed. Dörp-feld believes he has detected an outward and downward tilt of the stylobate blocks, rectified under the columns by a device which he seeks to identify with the scamilli impares of Vitruvius. See p. 328 of his article. I have nothing to say on this discovery.

⁴⁶ Each of these figures will be found under the appropriate member below.

⁴⁷ Grinnell restores thirteen flanking columns on her plan, op. cit., 38, but only twelve on her elevation, pl. XXXV, which she takes from Blouet, pl. 31. Her text ignores the problem.

⁴⁸ For a detail of epikranitis, which tempted me to restore a length of twelve columns and to remove the opisthodomos altogether, see below, p. 90 and pig. id.

⁴⁹ See also the remark in AE 1900, 116: Dörpfeld The treatment of the anta-returns I discuss below.

⁵⁰ The plan seems clearer than Dörpfeld's, but is misleading. We are not told who made it, but it seems to represent for the most part merely Dörpfeld's plan with the poros temple removed: and it has given the blocks of the south marble euthynteria the wrong colour. A patient excavation and survey of this temple is still, I hold, a necessity, even ifit brings no spectacular results.

⁵¹ It is columns that attract the tourist. ‘Lessuperbes colonnes de l'architecture hellénique … n'ont rien perdu de leur charme immédiat’ (Grèce, Baedeker 191, Préface). Stais also was irresistibly attracted by catalogues of standing columns. Witness his note 2 to col. 114 of AE 1900.

⁵² In his criticism, Mr. Lancaster will have taken this as an integral part of the ancient design.

⁵³ I know that where visible the opisthodomos foundations are nearly all four courses below the lowest marble course. But then they have all been levelled and tidied up, apparently in 1900 (AE 1900, 115). There are no published photographs of the excavations. Stais' erection only vaguely marks the temple's westward limits. He built his ‘ restoration (?) wall’ he tells us, between ‘the wall of the opisthodomos’, whatever that means, and the stylo-bate.

⁵⁴ Only for 3 cm. on each side of the vertical joints is the surface always smoothed in this and the two courses above.

^54a These are exact twins.

⁵⁵ The distance from the outer edge of the column to the outer edge of the stylobate.

⁵⁶ See above, p. 71, n. 17.

⁵⁷ The temples of Aphaia, Hephaistos, Poseidon (Paestum) and ‘Concord’ (Akragas) all oppose it.

⁵⁸ The fragment is small, but clearly from a corner. Orlandos is surely right to see the depression for a ceiling beam upon it (at a in op. cit., fig. 20); and as ceiling beams would here lie at right angles to the façade, the surviving portion of the block comes from the geison of a façade. We are therefore compelled to put it in Orlandos' position of op. cit., fig. 21.

⁵⁹ For the geison of the façades I follow AD, fig. 3.

⁶⁰ See above, p. 72.

⁶¹ See below, p. 101.

⁶² One also awaits a plan of the site that records the present position of all the surviving blocks now lying round the temple, and that numbers the more important. I had not the time or instruments for this.

⁶³ So Dörpfeld and Stais. While in my illustration I have followed them, in my own measurement I consistently fell short of their reckoning by as much as 0·15. As my measure was short (only 1·5 in length), I set aside my own reckoning. Only when I came to draw the pediment was I faced with the problem of the purlins.

⁶⁴ AD, fig. 13.

⁶⁵ I draw them in this position on fig. 2a, in firm lines. On the other hand, there was no centring over triglyphs at Aphaia, where each antefix surmounted a via. See Bagenal and Atkinson, Theory and Elements, fig. 86 (from Cockerell).

⁶⁶ III, 5, 12: the centre akroterion to be one eighth higher than the tympanum.

⁶⁷ How, with the fragmentary remains of the epikranits, could one tell which side faced the cella and which the peristyle ? I presume, from the marks on one side for the peristyle's marble ceiling beams and their correspondence with the geison blocks' marks. Here again I have to accept Orlandos' conclusion of 1917. At Rhamnous (see below, p. 104) the cella was probably higher than the peristyles.

⁶⁸ With some perversity I have strained in my drawing to make the inner shelf of the side wall's epikranitis equal to that of the entablature's. The latter was much wider, as AE, fig. 17 shows. I must leave this uncorrected, as the measures are too small for tidy correction.

⁶⁹ Vol. Ill, ch. I, plates IV and VIII.

⁷⁰ In this, presumably, the cella of the Hephaisteion resembles it. Dörpfeld asserts it as a rule that all semi-internal chambers in classical Greek buildings could be ceiled in marble, and that these included pronaoi and opisthodomoi with open colonnades on one side, but that genuine internal chambers must be roofed in wood. See AM 1883, 155. I can see no reason for this rule; and of course chambers elsewhere, 1 when without peristyles, were sometimes ceiled in marble, as, for instance, the famous example near Emporio on Santorin (Hiller, , Thera I, 306–7Google Scholar). I must, however, add that I am greatly indebted to this article of Dörpfeld on Philo's Arsenal. I have borrowed from it all my proportions for the rafters and cross-lathes of Greek temples; and on p. 151 it supports my nomenclature for ‘plinth courses’.

⁷¹ See above, n. 58.

⁷² See Shoe, Profiles of Greek Mouldings, 30 and pl. XXVI, 22.

⁷³ See above, nn. 27, 34.

⁷⁴ See below, p. 107.

⁷⁵ This would also align the anta-returns more artistically with the change from the isodomic anta-blocks to the pseudo-isodomic side walls, as we shall see at once from the probable lengths of the ‘plinths’.

⁷⁶ Two smaller fragments, whose length I measured as 0·77, have neither edge surviving. The curious can see the figures in Fabricius (AM 1884, loc. cit.). An epistyle block, now broken but perhaps from below the south end of the last sculptured frieze, has a dowel hole 1·19 from the base of its mitred end.

⁷⁷ See above, n. 35. I have not restored it in my picture of the Hephaisteion.

⁷⁸ See above, p. 77. I have scouted these proportions at the Hephaisteion. But Sounion as a whole is proportionately loftier.

⁷⁹ E.g. by Dinsmoor, 54, n. 120: ‘At Sounion, for instance, where these inward protruding jambs were planned, apparently projecting 0·285 m. inside the walls ( 1917, pp. 220–222), the sill-course is of exceptional width, 1·225 m. instead of 0·864–0·867 m. as in the Hephaisteion’.

⁸⁰ See above, p. 77 and n. 40.

⁸¹ It would, I suppose, be possible to give to a door 1·210 thick three lintel blocks of this width, and such a door could perhaps have rested on a sill-course 1·225 m. width (cf. n. 79 above). But I consider these figures a little close for an elegant building.

⁸² At the same time they are found in much later Ionic, such as Hermogenes' temples (Robertson, ² fig. 67).

⁸³ of intermediate, plans or views, the photograph of Rhamnous BSA I, pl. III, is of small use for my purposes. Frazer, 's plan (Pausanias's Description of Greece, II, 452Google Scholar—repeated in Frazer and Van Buren, Maps and Plans to Illustrate Pausanias, pl. XXII) is a mere repeat of Gandy's. One greatly regrets Gandy's failure to draw the site; for he tells us that in his time the steps were entire all round and the positions of the prostrate columns perfectly discernible. To-day less than one quarter of the stylobate survives, heavily incised by modern vandals. Gandy, moreover, found portions of ‘every part of the superstructure’ lying amongst the ruins, permitting a complete restoration of ceiling and roof. Gandy himself, as we shall see, took some fragments without recording the fact, and one metope, at least, seems to have found its way to Rome. Langlotz, in Scritti Hogara, 228, observes it could have come there ‘auf ähnliche Weise … wie die Fragmente vom Parthenon’, certainly on the building in Carrey's time, but now in the Vatican.

⁸⁴ Brought to my notice by Prof. Möbius.

⁸⁵ Loc. cit., 312. Clamps, as his plan shows (pls. IX–X), were used on the middle step at the corners only. Moreover, I saw another in a similar position on the lowest step. It fastened the N.E. corner block to its neighbour to the south. It measured 0·32 × o·08, and its axis was set 0·45 behind the face of the step. This reinforcement of the corners is an interesting sign of Greek logic and carefulness, and is repeated (cf. Orlandos, ibid.) at the corners of the secondary wall-base.

⁸⁶ Penrose observes, in his Athenian Architecture, when referring to the epistyle blocks of the Parthenon, that if all the stones ‘had been reduced to the length of an exact columniation, it would have been almost impossible to obtain a sufficient supply for the archi traves’. I believe that at Rhamnous also it was better to use the stones as they came.

⁸⁷ Orlandos, 311 and fig. 4.

⁸⁸ I followed Robertson's figure of 15·00 seeing that ca. 6·5 proved so accurate. Orlandos' plan is not meant to be accurate for differences below o·1. I have since found, on calculating from Gandy the lengths and number of orthostates, that I overshot the mark by about 0·1. This has not affected the dimensions of pronaos and cella, but has made the opisthodomos 0·1 too long.

⁸⁹ Robertson², 82, n. 3, gives this figure for the canonical Attic foot. It is a little large for the Hephaisteion, with its unit of 1·30 or four Attic feet. The figure given by Tilton for the Argive Heraion, which also observed the Attic foot, would agree more closely. The unit of the Heraion was apparently 1·633, or five Attic feet of 0·326. See Tilton ap. Waldstein, , The Argive Heraeum, I, 120.Google Scholar

⁹⁰ Op. cit., 44.

⁹¹ Robertson, figs. 39 and 43. The logical perfection of this Ionic plan is reached, in my opinion, with Hermogenes (Robertson, fig. 67), too frequently be littled as an architect. Lethaby's attack on him (Greek Buildings, 194) is unworthy of Lethaby.

⁹² See above, p. 81. This course, like the steps of Sounion, shows polished surfaces about the joints.

⁹³ See above, p. 95, note 86.

⁹⁴ This is the first place where I found Gandy seemingly unreliable. On pl. 11 he puts the second column over a joint on the long side of the temple.

⁹⁵ pl. 5.

⁹⁶ There are, however, difficulties. For the blocks are not always exactly 0·945, and one block still in situ has a roughening 0·54 instead of 0·47 long.

⁹⁷ E.g. his simplification of the drums, which I shall mention at once.

⁹⁸ Hesperia IX, 22.

⁹⁹ See above, pp. 71 and 82.

¹⁰⁰ See above, p. 71 and below, p. 111.

¹⁰¹ See above, p. 72 and 83.

¹⁰² Langlotz would place it, stylistically, late in the fifth century, but before its last decade. To my eyes it recalls in its manner the ‘summer-stricken’ art of the Bassae frieze.

¹⁰³ In both ‘a’ and ‘b’ of fig. 4 I have placed the ‘centre-block’ and change of joint very near one corner, at the left hand side. I do not know, of course, where it was, though I suspect from surviving blocks that it was near the left hand (‘dexter’) corner of each long side.

¹⁰⁴ See above, p. 95.

¹⁰⁵ Orlandos first pointed out (312–13 and fig. 5) that the front of the triglyph, as usual, was in the plane of the front of the epistyle, not behind it, as Gandy wrongly supposed.

¹⁰⁶ Here, however, arises a difficulty I have only recently observed. On Gandy pl. 5 and on my drawing, which follows it, the geison blocks, ceiling beams and coffers leave no space for the lower ends of the rafters. We may notice that Gandy has covered the side peristyles, as I have done, with the lowest ceiling beams available (0·208 in height; see below, p. 104), but even these give him no help here. Yet the rafters must have continued as far as the geison blocks, whose surface is designed to receive them in the customary manner. At present I Can offer no solution of the problem.

¹⁰⁷ On pls. 2 and 11 Gandy makes his corner blocks square, on a side of my length. But on pl. io they are oblong.

¹⁰⁸ Gandy places all the joints exactly over the centres of the columns. But in this temple the care of execution is not quite equal to the logic of the design. Moreover, the lower corners of the tympanum, as we see in the following note, have no direct relation to the columns, while its last 0·2 on either side may have formed part of the corner geison block.

¹⁰⁹ I may have misread my notes on the corner block of the geison; but I understand them to say that the upper slope of the tympanum dies into the upper surface of the geison 0·58 from the corner of the pediment, and that its last 0·2 metres form part of the corner block. This would give the whole a width of 5 × 1·8 plus 2 ·58 + 2 × 0·2, or 10·56—too wide a pediment.

¹¹⁰ In 1820. I am indebted to Mr. Ashmole, who showed me the record.

¹¹¹ Gandy pl. 12.

¹¹² Cockerell, Aegina and Bassae, Part II, pl. VII.

¹¹³ See Meurer, , JdI 1896, 132, fig. 20.Google Scholar Typologically, though not so primitive and uncoordinated as an example from Epidaurus, it shows the acanthus leafless organically placed than in an example from the Argive Heraion (Meurer, fig. 23). But I feel this is a matter of taste rather than date. (At Bassae there is no acanthus on the ridge tiles, but a very low acanthus calyx on the antefixes—Cockerell, pl. VII.) The tile itself was picked up by Herr Schiff ‘under the ruins of Rhamnous’ on Dörpfeld's voyage of 1895.

¹¹⁴ Shoe, 126 and pl. LX 12. The point of greatest projection is at the top.

¹¹⁵ Shoe, 6 and 126.

¹¹⁶ Orlandos was the first to assign the higher moulding, correctly, to an epikranitis and not, as Gandy had done, to the upper part of the frieze backers (Orlandos, fig. 5 and p. 315) just beneath the known epikranitis of the peristyles. At first sight the only other possible place for the larger moulding is the east peristyle, a part of the building specially treated by our architect. But the inner taenia of its epistyle corresponded to that in the other peristyles, as Orlandos shows (314), on all sides except its west side, which was the façade of the pronaos. One presumes that the epikranitis followed suit. But we know from the surviving internal mitre joint (Shoe, text, 126) that the heavier moulding covered at least two adjacent sides and must therefore have covered all four of the chamber where it was placed. None the less, I would welcome a clearance of the site of the sort I could not make myself last year.

¹¹⁷ See above, n. 83, and, for a difficulty, n. 106.

¹¹⁸ Architects are so much at variance here that I wish I had considered this feature more carefully. But I find on my return that my notes agree with Gandy (especially plate 3); Shoe gives the pronaos a floor level with the peristyle's, and the toichobate a wide, visible projection on both sides. But the toichobate was of fairly rough stones (cf. Orlandos, pls. IX–X) and barely meant to be seen even externally. In my illustration it is altogether invisible. Internally, in the pronaos, it was entirely covered by the floor slabs and secondary wall-base, which is about 0·7 wide.

¹¹⁹ The only parallel that comes to my mind is the Treasury of the Athenians at Delphi, where the ortho-states inside the building are disguised.

¹²⁰ For the pronaos, Shoe makes orthostates and backers together about 0·55. I believe with Orlandos, fig. 4, and Gandy that they were equal for the pronaos, though I hold they were unequal for the cella. Shoe sets her wall symmetrically on the wall-base. But Gandy, pls. I and III, shows this had an infinitesimal projection in the pronaos and none at all in the cella.

¹²¹ See below, p. 108.

¹²² See above, p. 102.

¹²³ With all Gandy's subsidiary mouldings, the base would total 10 ins. or 0·26 in height. But I would like to believe, on the analogy of Tegea, that the guilloche was the lowest moulding of all.

¹²⁴ Published by Dugas, and Clemmensen, , Aléa Athéna (Paris, 1924).Google Scholar Pl. XXII gives the elevation of the cella, pl. LXIV details of the base-moulding, pl. XCIIIB the photograph of a fragment. Though the temple is much larger than Rhamnous, the moulding is hardly larger, which disposes of objections that I am giving my cella a finicky base-moulding. At Tegea the cella is roughly 20 × 8 and the moulding only 0·295 in. height. At Rhamnous the moulding would be 0·16 to a cella of about 7 × 5. At the same time, the profiles of the mouldings differ considerably—at Tegea the Lesbian leaf is 0·9, the guilloche only 0·7 high. The reader can decide for himself from my fig. 5, where a and a (to be combined) are a quick tracing of Gandy, b a tracing of Clemmensen.

¹²⁵ My drawing makes them a shade too wide.

¹²⁶ This awkwardness does not show on my drawing. As already explained, I have made the opisthodomos about 0·1 too long.

¹²⁷ Incidentally I saw a block that fits the size of my lintel: 0·57 high, 0·74 wide and at least 1·32 long.

¹²⁸ Given both by Gandy and Orlandos.

¹²⁹ See e.g. Meineke, , Frag. Com. Grace, II, 80·6Google Scholar, Thieme, Quaestiones comicae ad Periclem pertinentes. Diss.