I use the following abbreviations in addition to those in general use:

A.A.G. Dinsmoor, W. B., The Architecture of Ancient Greece (1950).Google Scholar

A.M.S. Payne, H. and Young, G. M., Archaic Marble Sculpture from the Acropolis (1936).Google Scholar

The following works are cited by their authors' names alone:

Bohn, R., Die Propylaeen der Akropolis zu Athen (1882).Google Scholar

Bundgaard, J. A., Mnesicles (1957).Google Scholar

Heberdey, R., Altattische Porosskulptur (1919).Google Scholar

Koldewey, R. and Puchstein, O., Die griechischen Tempel in Unteritalien und Sicilien (1899).Google Scholar

Wiegand, T., Die archaische Porosarchitektur der Akropolis zu Athen (1904).Google Scholar

¹ Wiegand, 1–72. The metopes on one side at least were of Pentelic marble (pp. 10 ff.), and the sima (pp. 46 ff.) was of marble, again Pentelic according to Lepsius, (Griechische Marmorstudien 122).Google Scholar For the name, ‘H-fragments’, see Dinsmoor, AJA li (1947) 114.Google Scholar

² Payne (p. 38) dates this kore before 510 B.C. It is of Pentelic (Lepsius, 73).

³ Welter, , who restored the altar in AA 1939, 27–30Google Scholar, dated it to 512–511, and observed that we should not be misled by its rather Ionian inscription into dating it later. Klaffenbach, , in Kirchner, , Imagines ²11Google Scholar, is tempted to put it even earlier. Meritt's date, 497–496, has been generally rejected, and I myself follow the majority of scholars (cf. SEG x 318).

There are possibly older statues and structures of Pentelic (is the gorgoneion, Payne, pl. 1, generally assigned to the H-fragments, Pentelic?), but in my rather cursory searches I have not found them. Nor could I judge for myself whether the white marble of Weller's Propylon is really Pentelic. I am no mineralogist, and can recognise Penteliconly when it has rusted. But I am sure Wiegand's words are true (Wiegand, 12): ‘Je weiter aber unsere Kenntnis der attischen Marmorarten sich ausdehnt, desto mehr erkennt man, wieviel Vorsicht beim Bezeichnen geboten ist’—a conclusion reinforced by the paper of Herz, Norman and Pritchett, W. Kendrick on ‘Marble in Attic Epigraphy’, AJA lvii (1953) 71–83CrossRefGoogle Scholar, to which my attention has been drawn by Mr R. M. Cook. There seems atpresent, according to this paper, no short cut to recognising the quarry from the appearance of the marble. Herz and Pritchett agree with Lepsius that both ‘Hymettian’ and ‘Pentelic’ are found on either mountain, but do not share his optimism in assigning various grades of ‘Pentelic’ to the one or the other.

⁴ Wiegand, 59; Welter, , AM xlvii (1922) 69Google Scholar; Hill, B. H., AJA xvi (1912) 537 ff.Google Scholar

⁵ ‘… dass sie in überwiegender Anzahl nicht aus leichtem, gelblichen Poros, sondern aus schwerem, rötlichgelbem, hartem Kalkstein bestehen’ (Wiegand, P.5).

⁶ Wiegand, fig. 115; IG i² 925. I do not know why Dinsmoor calls the youth Λυσέας (AJA li (1947) 148).

⁷ Dinsmoor's Doric Foot is 32·7 centimetres long (A.A.G. 72 n. 1) or 32·6 centimetres (A.A.G. 195 n. 1). If, however, the Athenian foot is best measured in the stadium, it is nearly as the English 30·5 cms. and Wiegand's inner foundation must have been reckoned 114 feet long. (Buschor, 's ‘Hekatompedon’ at Samos was 33·5 metres long, according to AM 1930, Beil. 27Google Scholar). All the same, anything about 100 feet long could surely be called Hekatompedon by the crowd. Compare the use of χιλιοτάλαντος and Pentekontaetia. The cella of the Periclean Parthenon could be called Hekatompedon merely because it was almost 100 feet long, not because its forerunner was an exact Hekatompedon.

⁸ See his restored plan, AJA xxxvi (1932) 316. MrsHill, (Ancient City of Athens 136 f.)Google Scholar ignores this, surely the only tolerable restoration.

⁹ It had at each end a much deeper pronaos, with antae, and it did not have the thicker cross-walls of the Athenian temples. Indeed, it seemed probable to Doerpfeld that at Corinth the cross-cutting between the twocelias, being ‘much shallower and less sharply defined than any of the others, served … for some slight superstructure … probably a screen’. See Powell, B., AJA ix (1905)Google Scholar; and compare pl. 3. This would entail two continuous Doric colonnades along the whole interior—much the most likely design, it seems to me.

¹⁰ Wiegand, 119 f.; Lepsius, op. cit. 115 f. See above, n. 4.

¹¹ See Wiegand, 50, and also fig. 72, which shows the surviving stretch of euthynteria along the north side. An outline restoration, with euthynteria and stylobate only, appears in fig. 112. Dinsmoor, agrees (AJA li (1947) 117, n. 32)Google Scholar that ‘the outer pink foundation … served as the euthynteria course and supported the stylobate of the external columns’. The facade was 21·34 metres wide on a stylobate just over 1·5 metres deep. These are proportions very like those of the Heraion at Olympia—19 metres wide on a stylobate 1·4 metres deep.

¹² c. 600, because this is the probable date for the first peristyles even partly of stone. See now Searls, and Dinsmoor, , AJA xlix (1945) 73.Google Scholar

¹³ See, in particular, AE 1900, plate facing cols. 183–4.

¹⁴ As suggested by Rhomaios, in AE 1900, cols. 174 ff.Google Scholar

¹⁵ The west stylobate of Thermon is that of the High Archaic Temple (Rhomaios, loc. cit. col. 173). For Tegea, see AE 1952 1–31 (Rhomaios).

¹⁶ For the Basilica, see now Krauss, F. in Festschrift für Carl Weickert 99 ff.Google Scholar (fig. 1 gives 26 by 55 metres as the overall dimensions). Artemis at Corcyra was perhaps 23·45 by 48·9 metres on the euthynteria (Korkyra i, pl. 22).

¹⁷ AM lx–lxi (1935–1936) 72, fig. 8. He restores them as two lions.

¹⁸ The group of Herakles and Triton, as published in the standard books, now needs some minor corrections. See Hesperia viii (1939) 92 ff. (Broneer, ).Google Scholar

¹⁹ Dinsmoor, (AJA li (1947) 115 ff.)Google Scholar denies that Schuchhardt's reconstruction would fit the foundation. According to him, Schuchhardt obtained a temple 20·64 metres wide on the frieze, but should have obtained one 48 centimetres narrower, with ten metopes measured by Wiegand as each 1·115 metres wide, and eleven triglyphs each, according to Wiegand, 0–81 metres wide (Wiegand, fig. 24). Dinsmoor actually means 58 centimetres narrower. For 11 × 0·81 = 8·91, and 10 × 1·115 = 11·15, and 11·15 + 8·91 = 20·06, just 58 centimetres less than the width—20·64—which Dinsmoor supposes Schuchhardt to obtain. Actually, the difference is between 20·06 and 20·25, the width given in Schuchhardt, fig. 8. Since the metopes have nearly all gone, and since the temple was highly archaic, I do not find 19 centimetres an impossible difference. And at euthynteria level Wiegand and Schuchhardt differ even less. (For Dinsmoor's objection to Schuchhardt's pedimental slope, see below, p. 145.)

²⁰ For typical drove-work, unbroken striations of the flat chisel, on the H architecture, see, e.g., Wiegand, fig. 46 (from the soffit of a raking cornice—.cf. Wiegand, pl. 21), Schuchhardt, op. cit., pl. 16, where the work of the drove alternates with that of the small pick, or ‘punch’.

²¹ A fact stressed and perhaps exaggerated by Casson, S. in his Technique of Early Greek Sculpture (Oxford, 1933).Google Scholar

²² G. M. A. Richter, Archaic Attic Gravestones, fig. 73 ff. For the marks on this and another stele, see AJA xlvii (1943) 190, figs. 5–7 (Richter). Perhaps these stelai are not before 530, however.

²³ One remarkable instance of such a long, thin strip was evidently that pegged to the stone Kat. 1, shown in Schuchhardt, fig. 13.

²⁴ Payne puts Athena, from the pediment of the new temple, rather before 520. Wiegand, who first assembled the fragments of this ‘Pisistratean Temple’, dated it somewhere in the second half of the sixth century (Wiegand, 114), Dinsmoor, who accepted it as Wiegand left it, a little later than the Temple of Corinth, and therefore, I suppose, about 530 (A.A.G. 90). Apart from the pedimental sculptures, the best known fragments are the stretches of entablature built into the north wall of the Acropolis, of which Penrose, (Principles pl. 46)Google Scholar gives measured details.

²⁵ See, e.g., Dinsmoor, , A.A.G. 90Google Scholar, AJA li (1947) 117.

²⁶ It is rather difficult to assign the architrave block, Wiegand, fig. 3, to a peripteral building. Dinsmoor, saw this (AJA li (1947) 140, 143)Google Scholar, although he does not make his reasons clear. I shall have to return to this. See below, pp. 142 f.

²⁷ For all these details, see Hill, op. cit., pl. 8. For the moulded footing in the Theseum, see, e.g., Koch, , Studien zum Theseustempel pl. 44, 57 i.Google Scholar

²⁸ For pictures, see, e.g., Cavvadias and Kawerau, pl. 1. Photographs of the eighties have been constantly reprinted, e.g. in AJA xxxviii (1934) fig. 3; JdI li (1936) figs. 6 and 7.

²⁹ 3·146 at the ends, 2·146 at the sides (Hill, 547).

³⁰ Dinsmoor, 437–8. The sherd, Graef-Langlotz, ii 731, of c. 490) is much the latest. The other late sherds are of c. 510–500.

³¹ On p. 36 Kolbe argues that the technique of the lower part of the north wall and the ‘Cimonian’ wall is the same. But he seems to wish to prove by this not that the north wall is of the sixties, but that the lower part of the ‘Cimonian’ wall is of the seventies. I still think myself that the north wall dates from the ‘seventies, and is a good deal older than most of the ‘Cimonian’. It may, indeed, owe its jagged, ‘Cranaan’ plan not to its date but the nearness and steepness of the rock-face. Nor is it distinguished from the ‘Cimonian’ wall by the fact that it contains debris. One has only to think of Wiegand's epistyles embedded in the ‘Cimonian’. But remember that it displays die entablature of Wiegand's and most other scholar's ‘Pisistratean’ Temple (Wiegand, 118; see above, n. 23.) Surely this is an intentional memorial of the Persian barbarity, in which, to quote Picard, (L'Acropole, 12)Google Scholar, ‘le rationalisme esthétique des Athéniens avait mis de l'ordre dans les exercices de mortification patriotique imposés à la cité’. If so, ‘Pisistratean’ Temple and Parthenon will have been thrown down by the Persians.

Moreover, the north wall was more often seen and less of a task to build than any wall on the south. And finally Plutarch specifically connects the south wall alone with Cimon's victory of the Eurymedon and therefore, presumably, the middle sixties: (Plutarch, , Cimon 13·7Google Scholar). For the date, see Gomme, , Commentary on Thucydides i 408.Google Scholar I feel, then, morally certain that the north wall was built before the south.

³² See the excellent picture in Curtius, E., Stadtgeschichte von Athen 48Google Scholar, fig. 11.

³³ First Marble Parthenon, 2.81: 1. Corinth, 2.5: 1. Bassai, although of the later fifth century, is 2.5: 1. (I calculate the proportions of the First Marble Parthenon from Hill, 547, of the others from Dinsmoor, A.A.G. Appendix.)

³⁴ One thinks of Delphi and Ephesos. Above, I have argued for another, on the Doerpfeld Foundation. A similar history might help to explain Bassai. But the facts are still obscure. On p. 43 of A.A.G. Dinsmoor speaks of such an archaic temple as possible, on p. 155, n. 1 as certain. Kourouniotis, (AE (1910) 285)Google Scholar concludes that one existed and is represented not by the small ‘adyton’, but by the arrangement of the temple as a whole.

³⁵ Compare the cross-section, Ausgrabung pl. Θ and the two plans, pls. Δ′ and Ζ′.

³⁶ Compare Penrose, , Principles, pl. 2Google Scholar, Cavvadias, pl. Ζ′ and Stevens, G. P. in Hesperia Supplement iii (1940) fig. 32.Google Scholar Stevens in this paper, ‘The Setting of the Periclean Parthenon’, shows the rock-cuttings in great detail. I find his discussion on pp. 46 ff. of the earth-fills and buried walls rather less thorough—one reason why I felt I had to go into the rights and wrongs of the controversy between Dinsmoor and Kolbe.

³⁷ Not only would it have been a far greater labour to build out the platform even a few feet further, but also, no doubt, the Athenians up to 480 felt bound to leave proper circulation between the platform and the outer parapet of the Pelasgian Wall. Just sufficient was left at the platform's southwest corner.

³⁸ Note, e.g., Plutarch's portmanteau phrase, (Pericles 13.4). For a collection of ancient references, see Dinsmoor, AJA li (1947) 123 n. 74.Google Scholar

³⁹ Dinsmoor, , AJA li (1947) 124.Google Scholar Herington agrees with Dinsmoor, (Athena Parthenos 39–40).Google Scholar

⁴⁰ Dinsmoor, op. cit. fig. 7: ‘16·24 metres on the frieze’ (p. 141).

⁴¹ Dinsmoor, op. cit. 140. According to a very corrupt passage of Marcellinus' Life of Thucydides, Hippokleides son of Teisandros was archon when ‘Παναθήναια ἐτέθη’, and this, according to Jerome, was perhaps the Panathenaic Year 566–565; for Jerome runs ‘Ol. 53.3, Agon gymnicus, quern Panathenaeon vocant, actus’ (see Cadoux, T. J. in JHS lxviii (1948) 104Google Scholar). For myself, holding, as I do, to the dogmatic scepticism of Beloch, I denythat we know the true dates of any archons until nearly the end of the sixth century. So that on general grounds I consider 566 B.C. a worthless date. Davison, J. A., in ‘Notes on the Panathenaea’ in JHS lxxviii (1958) 27Google Scholar, seems to consider the style of the earliest Panathenaic vases to be about the best evidence for a date around 560. But at present we can date no vases of this time absolutely. Because of such vases a date of c. 560 has been proposed for the pedimental sculptures, Bluebeard and the rest, and accepted by most scholars. I have no experience in this field, but it seems to me that the heads and especially the beards of Bluebeard closely resemble those of the Panathenaic amphora in Halle (Beazley, , ABV 120Google Scholar, Development pl. 17.1); while Broneer's new head of Herakles reminds me even of Exekias (cf. FR pl. 131 and Hesperia vi (1937) 477). But even if vase-paintings and pediments were of 560 and so, perhaps, a generation later than Doerpfeld's Foundation, we need not be surprised. Hitches occur in building and tympana would be filled last of all.

⁴² This seems quite extraordinary construction, but I believe Wiegand, and feel bound to agree with him. How would water escape from the channels between this taenia and the metopes? The bottom of no metope seems to be preserved. But I suppose each could have had a projecting taenia, like the Sicyonian (for which see Coste-Messelière, La, Au Musée de Delphes 43Google Scholar, fig. 1, and pls. 1–3). If this came out to the level of the triglyph-face, as in the Sicyonian monopteros (op. cit., fig. 1), it could then shoot the rain-water down over che upper surface of the taenia of the architrave. In other stretches of the same architrave, however, the top was nearly flat, the construction nearly normal (Wiegand, fig. 6). The variations in the top surface are listed on Wiegand, 39 ff.

⁴³ I take this to be his reasoning, although I nowhere find it really explicit in his articles. On AJA li (1947) 115, however, he says that Schuchhardt's complete, encircling peristyle ‘is not only contrary to the evidence of the jogged joint on an existing corner architrave from one facade (fig. 8), but it forced Schuchhardt to eliminate all the flank architrave fragments, which are of forms that could not have been supported on columns and must have rested upon solid flank walls’. Dinsmoor, fig. 8 is a doctored version of Wiegand, fig. 3 (see fig. 6).So I think I have given Dinsmoor's argument. As for the flank architrave fragments, these are collected by Wiegand on pp. 39–43. I do not understand why Wiegand thinks they all come from flanking walls. We can prove that in this temple a fragment must do so only ifit is from the lower part of the architrave (and then how can we identify it?) and too short to stretch from column to column. For among the H-fragments, where triglyphs and sima were sometimes built up of irregular horizontal strips, could not the architrave be similarly constructed? In places, perhaps, it might even have blocks built in it at taenia-level, which did not meet one another at the centre of one regula. Blocks with architrave-mouldings but not originally of the architrave's full height are Wiegand, nos. 6, 7 (always a small block, just over 1 foot square), 11 (but is this really straight below?), 12, 13, 14, 16, and 20. Blocks with irregular side-joints were perhaps nos. 1 and 13. The original heights of these blocks varied from 30 to 52·5 centimetres. We shall return to this problem below, p. 155.

⁴⁴ Although rather ungraciously, for he believes that the normal metope-width was 117 centimetres, and that this penultimate metope, made deliberately narrower, happens to be 112 centimetres wide only ‘by a coincidence’ (op. cit., 142 n. 160).

⁴⁵ I am thinking of important Doric temples, not of such equivocal and perhaps secular buildings as the tristyle apsidal hall Wiegand, fig. 154, and what was possibly its successor, the north-west wing of Mnesikles' Propylaea. On the possible connexion of the two, see Bundgaard, 55 ff.

⁴⁶ Schuchhardt, fig. 8 and p. 87. The groups are Heberdey, nos. VIII and V (Triton and Bluebeard numbered as one).

⁴⁷ See, for instance, E. Lapalus, Le Fronton Sculpté en Grèce figs. 12 and 14. These groups are Heberdey, nos. VII and IX. All the poros groups, as Dinsmoor says (p. 146), were found close together.

⁴⁸ Dinsmoor, 145; Wiegand, fig. 24 (equal to the horizontal cornice); Heberdey, 142, makes it possibly less (‘at least 27 centimetres’).

⁴⁹ The step seems pretty ill-preserved. Ragged patches of it survive, attached to Herakles and Triton (Heberdey, V, E and F).

⁵⁰ Dickins', measurements (Catalogue i 78, 82).Google Scholar The figure to the left of Bluebeard, apparently peaceful, was envisaged by Dickins (p. 81) as standing, by Schuchhardt, more plausibly, as kneeling (Schuchhardt, 89, fig. 17).

⁵¹ Schuchhardt gets 15½; but one could run his monsters' snaky tails a little more into the corners. Some of these sculptures reached the tympanum frame. The heads of the snakes in the other pediment were sliced by the raking cornice; Heberdey, 104–5.

⁵² For a clear picture see Picard, C., L'Acropole pl. 75.Google Scholar

⁵³ On the Erechtheum, at the south-west corner of the North Porch: Hege, und Rodenwaldt, , Die Akropolis pl. 82.Google Scholar On the Propylaea, at the west end of the facade of the south-west wing: Bundgaard, fig. 7.

⁵⁴ I should wonder if they did, had not Dinsmoor seen on one of them the tongue pattern which comes on these metopes immediately below their crowning taenia (Wiegand, fig. 14a and b) and which is very plain on one of these metopes re-used for the Hekatompedon inscription (Wiegand, fig. 114; Kirchner, , Imagines ²20Google Scholar). Dinsmoor, writes (AJA li (1947) 143Google Scholar) that ‘two others may be restored by means of the painted pattern as 1·008 metres (Hekatompedon inscription, A) and 1·019 metres (one before the Propylon), these averaging 1·010 metres’. ‘He adds (p. 148) that the three still in place outside the Old Propylon “are complete apart from the retrimming occasioned by their new setting”.’

On pp. 142–3, Dinsmoor shows us that he needs metopes of Wiegand's height (1·37 metres—which, plus the top taenia, chiselled off, would be 1·5 metres)—and of varying widths, most of them 1·17 and 1·01, but a few of 1·12 and 1·06. The three metopes by the Propylon I measured as 1·30 high, and 1·00, 1·14 and 1·22 wide (3 feet 3½ inches, 3 feet 9 inches and 4 feet). These are awkward dimensions, and show either arbitrary trimming or a frieze not arranged like Dinsmoor's.

I sympathise, then, with Weller's note (p. 44): ‘I am informed that the opinion has been advanced viva voce that these slabs were originally metopes of the “old Athena temple”; how the irregular widths (cf. p. 47) are explained, I do not know. Besides, according to Penrose, (Principles of Athenian Architecture pl. 49)Google Scholar, the metopes of the old temple inserted in the northern wall of the Acropolis are 4·4 feet (= 1·3 metres) in height.’ Weller's measurements are ‘1·31 metres high, and from 1·23 to 1·01 metres long’—just a very little greater than mine, and quite incompatible, I think, with Dinsmoor's frieze.

In AJA li (1947) 118 n. 40, Dinsmoor says Weller expressed doubts on whether the Hekatompedon inscription used two metopes of the Old Temple. But Weiler doubted only the three slabs in situ beside the old Propylon.

⁵⁵ See p. 128, above. I could see no traces on it of the claw, only of the drove (see above pp. 131 f.). Unfortunately, my camera could not take this detailed work when I tried it in August 1956.

⁵⁶ Stevens, G. P. in Hesperia XV (1946) 78Google Scholar, fig. 4. These dimensions are repeated in Hill, I. T., Ancient City of Athens 143Google Scholar, and we are evidently not meant to believe her plan opposite, which makes them 13 by 13 metres.

⁵⁷ Bundgaard, 30 n. 21. He says that Stevens misunderstood Dinsmoor, . But in AJA li (1947) 126 n. 83Google Scholar, Dinsmoor writes about the Old Propylon that ‘we know neither the spacing of the columns, nor the distance between column and anta, nor the total width of the building (my discovery of foundation cuttings in 1910 having been incorrectly inter preted to the contrary)’. Bundgaard, however, says in his own note that in 1946 he himself carried out ‘a renewed exposure of the site’. He describes and illustrates the results on his pp. 41–44; p. 43 and fig. 22 make it clear that he only discovered a floating area of rough rock, on which it is unlikely that any walls stood. He thinks that the area is far enough north to exclude Stevens' north wall. But even this seems uncertain (p. 44). Perhaps this rock was outside the Propylon altogether. I cannot but recall Stevens' dictum (op. cit., 88), ‘The propylon … affords an illustration of the way the study of archaeology advances’.

⁵⁸ For all this, see Somers Clarke, and Engelbach, R., Ancient Egyptian Masonry esp. pp. 53 ff.Google Scholar

⁵⁹ For instance, if one raised the stylobate in the centre by a series of unequal steps (Vitruvius ‘scamilli impares’, as they are plausibly interpreted in Dinsmoor, A.A.G. fig. 59), it would greatly help in choosing the curve and determining the length of one's vertical module, if one drew an elevation beforehand. A plan or ‘paradeigma’ would have obviated confusion on the rather subtle east façade of the Theseum, where the axes of the columns are spaced, according to Koch, (Theseustempel 76 and fig. 70)Google Scholar, in a careful way. Take the mid-point (J) of the stylobate and that of the lower diameter in each of the three columns to one side of it (H, K, and L). Then JL: JK (1625: 1):: JK: KL (1·599: 1). The slight discrepancy even shows that the architect was using Fibonacci's figures (1, 2, 3, 5, 8, 13, etc.), for 1·625: 1:: 13: 8, and 1·599: 1:: 8:5. In actual modern measurements JL, JK and Kl are respectively 6·29, 3·87 and 2·42 metres (Koch, pl. 41) and I do not know in what ancient units of measurement they could have been worked out for the first time on the ground with movable pegs. Surely they determine the very length of the stylobate, before it is ever built! (For this, and other similar proportions, see my review of Koch, in Gnomon xxix (1957) 36–7.Google Scholar)

Any cornice block, especially one with the novel mouldings of the Propylaea (Bohn, , Propylaeen pl. 13Google Scholar), were better drawn beforehand than merely described or specified as the work went up. I find even a simple Greek cornice block almost impossible to describe. Therefore, I think, unlike Bundgaard, that the of IG II² 1666 A.48 must be some sort of a drawing:

There were to be others, of the anta-capitals (A. 90) and the tiles (B.34).

⁶⁰ Especially his brilliant use of dark stone under the wings to give the whole west façade a single continuous base, while at the same time the eye reads the four steps of the centre and the three steps of the wings each as single krepides, each of the correct height (one lower diameter) for the building above it. Elsewhere, too, he uses the dark stone to unify and yet articulate the building, thus showing he has those peculiarly architectural powers not, perhaps, exhibited in Scandinavia since the Renaissance Royal Palace at Stockholm. For the dark stone, see Shoes’ admirable paper in Hesperia Suppl. viii 341 ff.

⁶¹ At one point, I think, he explicitly decides for an unbroken gate-wall, but does not see how it wrecks his main theory. ‘It is hardly likely that the stair in the older propylon was treated as an independent feature, with special course heights which did not come within the gate-wall system till the threshold was reached’ (Bundgaard, Chapter I, 35).

⁶² For which see, e.g., Dinsmoor, in BCH xxxvii (1913) 57Google Scholar (opening of 2·4 by 3·2 metres).

⁶³ See p. 129 above. The graffito is ΛΥΣΙΑΣ ΚΑΛΟΣ. The metope is of dark marble, which may or may not be Hymettan (cf. Wiegand, 12). It does not seem to have had below the taenia the tongue-pattern found on other marble metopes among the H-fragments. But its taenia is nearly 22 centimetres high, and projects nearly 4 centimetres, which exactly agrees with their dimensions (Wiegand, fig. 14a); and its dressed face and top, like theirs, shows no trace of the claw.

⁶⁴ AJA li (1947) 117: ‘… all the “H” architecture and sculpture, which clearly belongs together without subdivision’.

⁶⁵ For the First Marble Parthenon, see Hill, , AJA xvi (1912) pl. 9.Google Scholar Hill restores it (i) on the analogy of the later Parthenon, (ii) because, in his view, the anta base which he describes on his pp. 552–3 can come only from an amphiprostyle temple. It projected from an end wall only about 0·50, and its projecting face, 1·822 metres long, corresponds to a column of the right width just under 2 metres. At the same time, one must notice that such an anta is unique. On one short side the moulding was cut off at a mitred joint, where the block met the cross-wall. But on the other short side, which is 120 centimetres long, and which marked the beginning of the cella-building's outer wall-face, the moulding continues unbroken. It should surely have been broken back a relatively short distance from the corner (here, perhaps, 40 or 50 centimetres) as is the similar moulding in the analogous position under the short return of the west anta of the Theseum (Koch, pl. 41). Presumably it was not, because the pronaos was prostyle—although, we notice, the anta-returns are relatively short on the prostyle façades of the Propylaea (Bundgaard, fig. 8) and the Temple of the Athenians at Delos, (Délos xii pl. 16).Google Scholar So perhaps the outer side of the anta equalled the face in width, as at Selinus G (Hulot et Fougères, Selinonte 251). But then, why does the moulded base-block not extend the whole width of the outer side? Would such asymmetric jointing on it be tolerable? Hill's plan (cf. my Fig. 9) gives the antae no break at all, and seems to make the moulding continue quite unbroken along the whole cella building. This, surely, would be tolerable only in the Ionic of the time. I am quite baffled by this detail of Hill's Parthenon.

⁶⁶ Well shown, for instance, at Bassai: Cockerell, , Aegina and Bassae (London, 1860)Google Scholar, Bassae, pl. 5: Durm, J., Baukunst der Griechen 3rd edition (Leipzig 1910 fig. 240a.Google Scholar

⁶⁷ See, for instance, the evidence for an extraordinary strip at one end of the raking sima (Schuchhardt, fig. 13 and pl. 1).

⁶⁸ Schuchhardt, fig. 13 (herringbone), fig. 7 (chequerboard); and Wiegand, pl. 9 (horizontal).

⁶⁹ Korkyra i pl. 24 (I do not know what is sima, what corona in the first phase—pl. 23).

⁷⁰ 26·0 centimetres, according to Schrader; 25, according to Payne.

⁷¹ As set up in the Museum at Delphi (cf. Robertson, , Greek and Roman Architecture pl. 2aGoogle Scholar).

⁷² Much depends on the restoration of the scrolls. Cockerell, (Aegina and Bassae, pl. 4)Google Scholar makes the akroterion much lower; Fiechter (in Furtwaengler, , Aegina, pls. 33, 49–51Google Scholar) rather higher.

⁷³ Dinsmoor puts it shortly after Thermon, which he dates to 620 (A.A.G. 51–2).

⁷⁴ For all this, see the first few pages of his article in AM lx–lxi (1935–1936). He calls the pediment with the ‘herringbone’ sima the ‘Blutengiebel’. For the ‘herringbones’ are really conventionalised lily-flowers laid on one side.

⁷⁵ For this jointing—in itself normal—see Schuchhardt, pls. 3, 7, 5.4 and 6.5, the three clearest in stances (= Kat. 14, 15 and 21)—all from the herringbone sima. Schuchhardt seems to think that the intended unit of length was just over 70 centi metres, that this was possibly observed by the clay pantiles which, in his view, composed the bulk of the roof, and that discrepancies of jointing near the edges could be adjusted under the cover-tiles (easier, this, with joints down the roof-slope than with those across it). See Schuchhardt, 104, which I am not sure that I have perfectly understood.

⁷⁶ P. 115. ‘Schuchhardt catalogued the 107 fragments of marble sima known to Wiegand, and added 109 new fragments to the list, a number which Züchner in the same year increased by 62 pieces, while Broneer found a few others in a well on the north slope of the Acropolis in 1937, so that we are now dealing with a grand total of about 285 pieces., But Dinsmoor, while possessing these 285 to Schuchhardt's 216, does not himself re-examine them as evidence for the width of the roof.

⁷⁷ Wiegand (p. 12) calls them ‘Kalkstein’ (= calcaire), Dinsmoor, calls them ‘poros’ (AJA li (1947) 148).Google Scholar Alas! when in Athens I did not find these fragments which all seem to be small.

⁷⁸ For instance, Dinsmoor, 's plan (AJA xxxvi (1932) 316)Google Scholar makes them only two-thirds the diameter of the outer. But then it seems to wish to make them Ionic—unnecessarily ?

⁷⁹ The Temple of Zeus at Olympia affords a very good example, where the pronaos columns are about five-sixths of the peristylar, and the entablatures are nearly equal (Olympia, Tafelband I, T. IX and X).

⁸⁰ This is well shown on the Theseum (Koch, pl. 6) and, in an analogous position to ours, on the north-east corner of the Propylaea (Bohn, pl. 6).

⁸¹ Here the regula was much wider than the upper diameter and nearly as wide as the lower, and the joint between the architrave-blocks on the corner came nearly half way along the regula. See Coste-Messelière, La, Au Musée de Delphes figs, 1, 4Google Scholar, which the reader should study before condemning my sketch, fig. 14, out of hand.

The small Sicyonian architrave, like the large architraves of Paestum, needs no ‘backer’ and is only one block deep. Among the fragments of H architrave-blocks measured by Wiegand (pp. 3–4), none seems to be thicker than 31 centimetres.

⁸² In a normal prostyle porch, the axis of the last triglyph is the length of one good architrave-block from the centre of gravity of the anta (see, e.g., Délos xii pl. 16). But could it be here, especially if the last triglyph and the corner-column had the same axis? Wiegand's block is 3·84 metres long, which should mean that the front of the anta was just over 3¼ metres from the axis of the column. But the Doerpfeld Foundation, as shown in Wiegand, fig. 72, suggests that anta-front and column-axis were only 2¾ metres apart. The foundation, very fragmentary at the corners of the cella-building, could perhaps be deceptive. Or perhaps in my sketch I have made the anta too narrow on the outside. Give it an outer face as wide as its front to the column, do not bother to centre the third triglyph over it, as in my sketch, and on the inside give the anta the slightest possible projection from the cross-wall, and perhaps one could then fit my architecture more easily on to the Doerpfeld Foundation. See my plans, Fig. 15, where (A) gives the restoration that I like, and that I have shown in Fig. 13, (B) the restoration, with narrow pronaos, which better suits the foundations of the corner as given by Wiegand.