You are on page 1of 22

Frames designed by Leonardo with short pieces. An analytical approach.

Jos Snchez & Flix Escrig. Seville University, School of Architecture, Avda. Reina Mercedes 2, Seville, Spain
josess@us.es felix@us.es http://www.performance-starbooks.com www.starbooks.es

KEYWORDS Historic structures, rapidly erected structures, reciprocal structures, wood structures, steel structures ABSTRACT. This paper deals with the analysis with actual mathematical tools of some structural proposals done by Leonardo in his notebooks. At difference of other subjects such as the Fly of Birds or Treatise of Painting, he never made his sketches to establish a coherent treatise on structures although he consumed a lot of pages analyzing constructional patterns to build bridges, roofs and floors a well vas beams, strings and mechanisms. May be he never constructed his proposals, but them are designed as if it would necessary to define minimal details. A particular subject contained in his notebooks are concerning to great span structures built with short elements. We try to proof the validity of these proposals. INTRODUCTION: It is now commonplace to refer to Leonardo every time we touch upon a theme related to art, science or technology, because we can always find some new aspect of his manuscripts which has not been studied or exploited previously. This is explained by the fact that his manuscripts were not published until five centuries later (Ref. 17). His observations about designing of structures are less well known and also the least studied. Nowadays, his most important contributions in this sense are contained in the Madrid Codex I and II, (Ref. 11) and some other more general entries in the Atlanticus Codex (Ref.10). His other notebooks do not deal with these questions. 1. Designing structures to span great distances with short pieces. Leonardo, in his drawings, manifested an obsessive search for solutions to the problem of how to span great distances with short pieces. In his notebooks, we find reiterative diagrams which seem to be searching for variants of frames which could constitute structures, mainly roofs, floors and bridges, as to cover large distances by using manageable pieces which could be easily supplied. He probably did not go beyond drawing his inventions and did not put them to the test experimentally but, nevertheless, he resolved all the details as if he was really going to build them. We are going to develop some of these applications with the light of the knowledge we now possess. It is well known that Leonardo was familiarized with a lot of previous studies relating to inventions and structural solutions, contained in numerous old and contemporary treatises (Ref. 3). He did his apprenticeship in Siena, Milan and Florence at the time of the great constructions. In Siena, he was in contact with the school of engineering, whose maximum exponent, Giorgio Martini, wrote and published an excellent treatise entitled Trattato di Architettura, Ingegneria e Arte Militare (Ref. 1

12) which, through its magnificent illustrations, helped to spread the current state of the art. He also worked on the Duomo di Milano (Milan Cathedral) and on an ingenious design to lift up the cathedrals dome, which has been preserved (Figure 1) (Refs. 10 and 13). In Florence, he was impressed by various Brunelleschis constructions, especially the great dome of Saint Mary of the Flowers. He was aware of Brunelleschis inventions for the construction of this dome which, although they were never published, circulated as working materials. He also knew Albertis Ten Books of Architecture, almost an obligatory reference which, although not very detailed in terms of specific solutions, were very suggestive on a general level. Perhaps he did not actually know other medieval references such as those contained in Villard de Honnecourts drawings in his Livre de Portraiture (Ref. 18), but it is known that ideas taken from such works already comprised part of the repertory of conventional solutions for the architects and engineers of the time (Figure 2).

Figure 1. Studies for the construction of the dome of Milan Cathedral. Sheet 310r-b of the Codex Atlanticus. Figure 2 Page 23 of the Villard de Honnecourt Notebook.

It is not surprising, therefore, that Leonardo attempted variations on themes which he could consult and be familiar with, and that he was not entirely original when it came to conceiving his solutions. His intention, in many cases, was to re-design and improve an existing idea. He expressed himself much better in drawings than in actual construction, at least as far as his engineering and architectural proposals were concerned. Maybe, given the surprising degree of precision and detail of specific solutions, the direct commissions which he tried to resolve in his drawings have not reached us, and his proposals were not just dilettantism but of practical application. In any case, we are going to study various examples of applications of this inventive capacity (Refs. 5 and 7). 2. Bridge to be assembled quickly with unprepared trunks. In the Codex Atlanticus (sheet number 22r) (Ref. 10) we can find a bridge built with logs with which it can be supposed that Leonardo intended to provide a simple, economical and rapid solution to overcoming such obstacles as streams and river beds (Figure 3).

Figure 3. Drawing on sheet 69r of the Codex Atlanticus. In this sketch we can appreciate the details of erecting the bridge: a) Preparing the flat aligned rods to constitute main arches. b) Connecting by means of transverse bars both bundles c) Designing tripods to lift arches d) Tying the grid to make rigid the platform d) Placing the assembly ready to be lift f) Putting the transverse poles to immobilize the erected bridge. The text below the drawing correspond to sheet 57v and it refers to the way of building the bridge with an indicative line: Questi legni vogliamo essere sottili a guisa di lance, acci che con facilita si possino elevare dal piano dellacqua e fermarsi, en alto tanto che in quattro o cinque luoghi smbastica il ponte. Poi nel luogo delle lance metti legni grossi e fornisci di tessere il ponte di due braccia in due. These woods want to be thin like a lance, so that easily can be raised and be strunken once fixed in four or five points. Then put in deep wood spears to suport the bridge with two arms into two arms.

What is really complicated is how to put this combination of pieces together in an orderly fashion so that it works, and that is precisely what is specified in various sequences of his drawing in the figure 3. On the left of the drawing and from top to bottom, first can be seen the alignment of longitudinal logs and their crossbeams positioned in the mouths of the extremes before fixing them in place. This entire flat system is secured with ropes in order to be able to push it towards the river and be supported by provisional tripods inside it. Then comes the raising and when the overall is raised the transverse log of sections are being pushed inwards. When the bridge is correctly lifted 3

the provisional tripods can be withdrawn by working from the inside outwards. Only someone who was thinking of constructing it could approach it in this way. A really impressive lesson of technique is contained in this brief page, and no sketch is idle or without purpose. The skill as a draughtsman is obvious in the facility with which he makes it perfectly evident which pieces cross and which pieces pass in front or go behind, without any confusion, and what is more it seems to be quite clear that it does not need to be secured once the assembly has been completed and that, with correct horizontal support no piece can now move out of position (Figure 4).

Figure 4. Different steps to erect Leonardos bridge according to description of figure 4. Note that bars are of the same length and that the assembly is stable in each step hanging from tripods till we cross the transversal elements. No ties are needed except the initial screws or ropes where the bars twist because otherwise the whole is a mechanism. These screws can be removed when the erection is finished.

Figure 5. Geometrical evolution of erecting according to Figure 4.

In the figure 5 we show the rising of the whole with the bar position changing. Although Figure 3 contains the most advanced project, Leonardo made several sketches in other notebooks before arriving at the solution. (Figures 6 to 8) (Refs. 10 and 11).

Figure 6. Bridge drawn on sheet 45r of the Madrid Notebook I. We can see that instead as single transverse pole he uses two of them to make possible increasing the rise of the bridge

Figure 7 Bridge drawn on sheet 46r of the Madrid Notebook I. It is useful for short distances but is weaker than other proposals and it needs longer pieces.

Figure 8. Bridge drawn on sheet 183r of the Codex Atlanticus. The text below the bridge explains that the height of the arch depends of the size of transverse rods.

The difficulties in arriving at these solutions made him prefer to develop the proposal in Figure 3 in greater detail. He never made an analysis of the structural behaviour, in part because he did not know much about vector calculus and also because this was not the matter he was most concerned with. On other occasions he was more concerned with these analytical aspects, as can be seen in Refs. 4,15 and 16. 5

The analysis which we carried out ran into certain difficulties in creating and testing the models of these simply-supported connections and passing bars which we have only been able to overcome by means of a second-order analysis in order to reach coherent solutions. By means of the SAP2000 programme we started off from Figure 4, where the transverse logs have been substituted by bars with an articulated connection. The analytical plan can be seen in the figure 9 with 5m-long bars, self weight and live loads and sections as detailed. The total height is 5.63 m. and the total span is 15.30 m. The loads considered in any case are self weight as dead load and non symmetrical loads a s shown in the figure 9. In Figure 10, a diagram of bending moments with loads applied at the nodes, and two extremes articulated at the foundations is shown, while in Figure 11 we show the axial forces in this case.

Figure 9. Structural plan of a Leonardos bridge.

Figure 10. Diagram of bending moments due to dead loads applied at the nodes and ends articulated at the foundations.

Figure 11. Diagram of axial forces due to dead loads applied at the nodes and ends articulated at the foundations.

Figure 12. Diagram of bending moments due to non symmetric live loads applied at the nodes and ends articulated at the foundations.

Figure 13. Diagram of axial forces due to non symmetric live loads applied at the nodes and ends articulated at the foundations.

Figures 12 and 13 show bending moments and axial forces for non symmetrical live loads.

Figure 14. Diagram of displacements due to non symmetric live loads applied at the nodes and ends articulated at the foundations.

In figure 14 we show displacements for non symmetrical loads that are 48 mm. in horizontal and 11 mm in vertical. We will consider now the possibility of sliding foundations the results are quite different, but it works efficiently and it is more similar to Leonardos proposal Figures 15 to 18).

Figure 15. Diagram of bending moments due to symmetric dead loads applied at the nodes and sliding foundations.

Figure 16. Diagram of axial forces due to symmetric dead loads applied at the nodes and sliding foundations.

Figure 17. Diagram of bending moments due to non symmetric live loads applied at the nodes and sliding foundations.

Figure 18. Diagram of axial forces due to non symmetric live loads applied at the nodes and sliding foundations.

With respect to displacements it occurs with 20 cm of vertical deflection and 20 cm of horizontal displacement on the foot to dead loads, and 75 cm in vertical displacement and 145 cm for 8

displacement of the foot in non symmetric loads. This means that sliding foundations produce excessive displacements. We could think that Leonardo invented the system without prior references but it seems highly improbable given that the accounts of travellers, especially those coming from the East, stimulated the imagination of European architects, as is well documented. It is also a fact, however, that the Rainbow Bridge itself, of the Song Dynasty, did not have any known sequel, even in the same country, and that by the 15th century the bridge now no longer existed except in the silk scroll painted by Zhang Zeduan, a 12th century artist, in one of the most impressive panoramas of Chinese art, which is kept in Beijing museum under the name Qingming Festival along the Pien River (Figure 19), (Ref. 20). It is difficult to think that he did not, given the great similarity of the solution, although he put more emphasis on constructive solutions than on the idea itself, which he seems to recognise as coming from an outside source.

Figure 19. Bridge from the Song dynasty from the Qingming Festival along the Pien River.

Figure 20. Nova Project directed by Bashar Altabba for American television which attempts to recreate the Rainbow Bridge.

Recent attempts to reproduce Leonardos bridge or its oriental precedent did not put their trust in Leonardos solution without any fixing and have preferred to ensure stability with screws where the bars cross (Figure 20) which was actually built using the same idea (Ref. 1). 3. Roofs and Floors. Another problem which Leonardo put a great deal of effort into solving was the analysis of roofs and floors, resolved with short pieces, shorter than the opening to be covered. As in the previous case, Villard de Honnecourt had represented solutions for this problem in his notebook (Fig. 2) (Ref. 18). In the future, this solution would be called Serlian floor in spite of his being merely an author of disclosure. It was really Leonardo who first posed the problem in absolutely general terms. His first plans were very elementary and were based on known solutions (Fig 21).

Figure 21. Drawings on sheet 50v of Madrid Notebook I.

Current researchers and designers have coined the term Reciprocal Structures for those which sustain themselves in simple or generalised circle form (Ref. 14). We are going to compare de behaviour of the model shown in the figure 22 with a conventional clamped connection. Loads considered are self weight and live load 150 Kg/sqm for dimensions considered in the figure 23. With these properties the bending moments for hinged joints are shown in figure 18 while figure 19 shows the results for clamped ends.

Figure 22. The wooden model made of Leonardos model.

Figure 23. Dimensioning the figure 22.

We can see a great coincidence between two structures respect bending moments, but not with respect deflection, because in the firs case we have 3.60 mm. and 3.00 mm. in the second (Figures 24 and 25). 10

Figure 24. Resulting bending moments distribution for hinged ends. Figure 25. Resulting bending moments for clamped ends.

Figure 26. Codex Atlanticus sheet 889v where we can see different designs of reciprocal frames. The text contains only a description of number of pieces necessary to build each grid and the length to be done them.

Figure 26 shows the complexity in which Leonardo moved. It can be discussed if these drawings are structural or pure geometries, but the detail with they are explained demonstrate the constructive intention. From a structural point of view it seems impossible that these design patterns drawn by Leonardo could function. There appear to be too many false supports for the transmission to work. But the physical and analytical models confirm that not only do they work, but also those they do so with great efficiency. In the next figures we will show examples of these that have been built. It is truth that these assemblies only work correctly if the perimetral extremes are fixed and we will consider so in our analysis. 11

In Figure 27 we can see the model which develops drawing A1 of Figure 26. It is a matter of dividing the current bar type into three parts and supporting the pieces on each other as indicated. If we make a indentation in the extreme of the bars, as can be seen in Figure 27, the final development will be flat. Nevertheless, if the pieces support each other taking their combined thickness into account, the final form will be curved in two directions, as will be seen in an analysis which will be made below. The analytical approach to the structures based on these patterns has been undertaken with the simplifications necessary for geometrical definition. Namely: - The bars are 3m long and the support points are indicated at 1m. - Sections are #300x150x6mm - We consider the self-weight loads to be applied at the nodes and live loads of 150 Kg/sqm. - Bars are hinged at the ends. - The model is therefore flat. In the figure 28 we can see the bending moments diagram for self weight and live loads considered above.

Figure 27 Model of the pattern drawn by Leonardo in Figure 26 A1. Figure 28 Resulting bending moments distribution for self weight and live loads for hinged ends.

Figure 29. Diagram of the bending moments for clamped ends. Figure 30. Diagram of the torsional moments for clamped ends.

If we compare the results of a floor such as that in Figure 27 with a conventional one, with all joining points rigidly connected, and the same overall dimensions, the same extreme supports and the same total weights, we obtain the results shown in the figure 29. The difference in bending 12

moments is 9% while it is 10 % in deflections (3.3 and 3.0 cm.). The only problem in the clamped extremes is that it introduces torsional moments a shown in the figure 30. In Figure 31 another pattern which Leonardo drew is shown, this time drawn much less clearly on the same sheet indicated as A2 in Figure 26. It is a tessellation made up exclusively of squares of two different sizes.

Figure 31 Model of the pattern drawn by Leonardo in Figure 26 A2. Figure 32 Resulting bending moments distribution for self weight and live loads for hinged ends.

If we compare the results of a floor such as that in Figure 32 with a conventional one, with all connexion rigidly joined, and the same overall dimensions, the same extreme supports and the same total weights, we obtain the results shown in the figure 33. The difference in bending moments is 30% while it is 13 % in deflections (1.8 and 1.47 cm.). The only problem in the clamped extremes is that it introduces torsional moments a shown in the figure 34.

Figure 33. Diagram of the bending moments for clamped ends. Figure 34. Diagram of the torsional moments for clamped ends.

In Figure 35 another pattern which Leonardo drew is shown, this time drawn much less clearly yet on the same sheet indicated as A3 in Figure 26. It is a tessellation made up exclusively of squares of the same size. 13

Figure 35 Model of the pattern drawn by Leonardo in Figure 26 A3. Figure 36 Resulting bending moments distribution for self weight and live loads for hinged ends.

If we compare the results of a floor such as that in Figure 36 with a conventional one, with all connexion rigidly joined, and the same overall dimensions, the same extreme supports and the same total weights, we obtain the results shown in the figure 37. The difference in bending moments is 130% while it is 68 % in deflections (3.41 and 2.03 cm.). This means that the reciprocal frame in this case is less apropiate that the conventional one. The torsional moments introduced in this case do not penalize the results (Figure 38).

Figure 37. Diagram of the bending moments for clamped ends. Figure 38. Diagram of the torsional moments for clamped ends.

With regard to the pattern obtained with triangular grids (Figure 26 B1) the conclusions obtained are similar. In Figure 39 a model can be seen and Figure 40 shows a scale distribution of the bending moments due to loads at nodes for a flat model with articulated bars.

14

Figure 39. Model corresponding with the pattern B1 in Figure 20. Figure 40. Results obtained in bending moments for the flat model with the same simplifications as made in the models of Figures 27, 31 and 35 for articulated bars.

If we compare the results of a floor such as that in Figure 40 with a conventional triangular grid with hinged extremes we obtain the bending moments shown in the figure 41 and the torsional moments of figure 42. The difference in bending moments is 10%, while it is 10 % also in deflections. In this case, the results of the layout in hexagons is less efficient than the layout in rectangles, for the same total weight and for a flat structure.

Figure 41. Diagram of the bending moments for clamped ends. Figure 42. Diagram of the torsional moments for clamped ends.

4. Considerations about the thickness of the bars. If we consider that pieces do not connect in line but are supported on the upper face and that, therefore, take on the form corresponding to a curved surface, such as can be appreciated in the models in figures 43 to 46 corresponding to flat models in figures 27, 31, 35 and 39. Then the analytical model changes and the results also change radically especially if we articulate the edge nodes and we do not allow them any displacement.

15

Figure 43. Model corresponding to drawing A1 of the figure 26. Figure 44. Model corresponding to drawing A2 of the figure 26.

Figure 45. Model corresponding to drawing A3 of the figure 26. Figure 46. Model corresponding to drawing B1 of the figure 26.

. In these cases we are going to compare only the effectiveness of pattern A1 of the figure 26. If we compare the results of a spatial model such as the one in Figure 47 with a flat one in the figure 48 we find that there are substantial differences, now in favour of Leonardos model, as axial forces are introduced in bars for the transmission of loads. Pipes are in these cases D100.6 and loads are ten times the self weight (10xDead Loads) applied along the bar length.

Figure 47. Curved model due to the thickness of bars that are leaned one on the other only preventing the sliding. Figure 48. Flat model with bars rigidly connected.

In Figures 49 and 50 we can see the bending moments out of plane M33and in plane M22, and comparatively, the difference in the values of the bending moments out of plane in the figure 51with the model of the figure 48. While in the first the maximum bending moment is 0.30 Tonxm, in the second, 3.17 Tonxm is reached for self weight amplified ten times. 16

Figure 49. Bending moments M33 for Leonardos spatial model.

Figure 50. Bending moments M22 for Leonardos spatial model.

17

Figure 51. Bending moments for the flat model.

In Figure 52 the results of the axial forces can be seen for the spatial model and it is not compared with the flat model because this model does not generate these forces.

Figure 52. Axial forces for Leonardos spatial model.

18

5. Other Leonardo models. In the same drawing in Figure 26, and in the sketches indicated as C1 and C2, Leonardo proposes constructing an extremely rigid platform based on the same principles as his bridges but with curved bars. This platform is constructed in the model in Figure 53, and although it appears to be of an elementary geometry it does, in fact, have a certain complexity.

Figure 53. Rigid platform composed of short curved bars, as can be seen in drawing C1 in Figure 26.

The question here, the same as in the other cases seen previously, is whether this construction has structural efficiency compared with the conventional designs which are similar to it in form. As it is very difficult to determine what is the similar conventional structure we only make an approximate calculus of Leonardos proposal.

Figure 54. Unit of Leonardos platform that we are going to treat as a flat structure.

Let the structure in Figure 54 be the one which can resemble an equivalent flat model and the two analytical models can be seen in Figure 55. Note that in some nodes there are continuous bars which pass and others which articulate at that node.

Figure 55. Modelling the truss of Leonardos platform C1 of the figure 20

The bar sections considered are 100.th.6 in size and the loads are, in the case of the hypothesis, self weight and 200Kg vertical load at each node. One extreme of the beam has sliding supports. The total length of the trusses is 6 m. Stresses are shown in figures 56 and 57. 19

Figure 56. Axial forces

Figure 57. Bending moments

The platform in Figure 53, resolved with short curved bars, behave with great efficiency, as we have seen, and provide a useful model for structural applications. 6. Conclusions. All of the design patterns studied in this paper are taken from drawings of Leonardos and it is noticeable that, in spite of the numerous studies which have recently developed the concepts of reciprocality or mutual support, there have been very few practical applications.

Figure 58. Drawing of Serlio First Book of Architecture Figure 59. Design analyzed by John Wallis for the Seldonian Theatre in 1689

We can cite as precedents drawings by Serlio (Ref .19), which probably knew several precedents not only by Leonardo (Figure 58) and the Seldoniam theatre roof in Oxford from Christopher Wren an Robert Wallis, never build, that we know because the precise analysis made by the mathematician John Wallis (Figure 59) (Ref. 21). Other interesting design based in the same principles of reciprocity is shown in the Frezier treatise (Figure 60)(Ref. 8 and 9)

20

Fig 60. Frezier sheet of a flat roof with all equal pieces.

In some of our architectural applications we have made use of this knowledge of Leonardos in order to define modular structures, that is, whole structures made up of small pieces combined to form large structures. The contents of Reference 6 (Figure 61) can serve as example

Figure 61 Modular systems which were used for the construction of the training pavilion in the Huelva sports complex by Escrig and lvarez in 1988, and previously in the Pavilion of Extremadura of the Expo 92 in Seville by Escrig and Curbelo. It is a degeneration of the proposal in Figure 35 or even 58.

We have tried to explain and account for the fact that structural applications are possible and that their versatility is immense. In order to expand this information from a conceptual point of view, you are recommended to look at Reference 2. Maybe Leonardo was a frustrated researcher who hardly constructed any of his inventions, although this affirmation is doubtful in view of the perfection with which he worked on the details of his 21

drawings. But what is certain is that his work was lost until the early 1950s, and it is now that we can take advantage of much of his knowledge. This is the main aim of this paper. 7. References

1. Bashar Altabba P.E."Re-creating the Rainbow Bridge" ASCE Civil Engineering nr.5 May 2000. http://www.pbs.org/wgbh/nova/lostempires/china/builds.html 2. Baverel, O., Nooshin, H. Nexorades Based on Regular Polyedra. Nexus Network Journal 2007. Kim Williams Books, Turin. 2007. Pp 281-298 ISSN 1522-4600. 3. Duhem, Pierre Etudes sur Lonard de Vinci. Ceux quil a lus et ceux qui lont lu Librairie Scientifique A. Hermann. Paris 1906. 4. Escrig, F., Snchez, J. La geometra de las estructuras I. STAR 12. Sevilla. 2005. 5. Escrig, F., Snchez, J. Adaptable Leonardo. International Conference on Adaptable Building Structures. Eindhoven. July 2006. ISBN 90-72152-03-4.pp. 5.28 a 5.32.
6. Escrig, F.; P. Valcrcel, J. "Modular Space Frame Structures". Space Estructures. Ed. Parker. Thomas Telford, London 1993. ISBN: 07277-1968-8. 7. Escrig, F., Snchez,J. Una aproximacin analtica a las mallas recprocas diseadas por Leonardo Informes de la Construccin Vol 62,518, 5.14 abril 2010 ISSN 0020.0883. 8. Frzier, A.-F., 1737 (ed. 1980): La thorie et la pratique de la coupe des pierres et des bois pour la construction des votes et autres parties des btiments civils et militaires. Nogentle-Roy : Jacques Laget L.A.M.E 9. Fleury, Francois. Evaluation of the Perpendicular Flat Vault Inventors Intuitions through.Large Scale Instrumented Testing. Proceedings of the Third International Congress on Construction History, Cottbus, May 2009

10. Leonardo Da Vinci . Codex Atlanticus. Ambrosiana Library. Miln. 11. Leonardo Da Vinci. Madrid Notebooks I and II. National Library. Madrid.
12. Martini, Francesco di Giorgio. Trattati di Architettura Ingeneria e Arte Militare Milano Il Polifilo, 1967

13. Pedretti, Carlo. Leonardo architetto. Electa. Milan 1988. ISBN 8843525190. 14. Popovich, O. Reciprocal Frame Architecture Architectural Press. Amsterdam 2008. ISBN 978-0-7506-8263-3 15. P. Valcarcel,J. Escrig, F. "Pioneering in expandable structures: The Madrid I Notebook by
Leonardo da Vinci". Bulletin of the IASS. Vol 35. April 1994, n1 114. pp. 33-45. Madrid. ISSN 0304-3622. 16. P. Valcrcel,J; Escrig, F. "Un primer planteamiento de estructuras desplegables. el Cdigo I de Madrid de Leonardo da Vinci". Boletn Acadmico de la ETSA, n1 10. La Corua, 1989. Pp. 13-20. ISSN 0213-3474

17. Truesdell,C. Essays in the History of Mechanics. New York: Springer-Verlag, 1968 18. Villard de Honnecourt. Cuaderno Ed. Akal 1991. ISBN 84-7600-670-5. Madrid 19. Serlio, Sebastiano Libro Primo dArchitetura di Sebastiano Serlio Bolognese Venetia 1566. 20. Silk scroll in the Beijing Museum. Qingming Festival along the Pien River
http://www.huangshantour.com/english/river.htm

21. Taylor, Paul H. John Wallis and the Roof of the Sheldonian Theatre - Structural Engineering in Oxford in the 17th Century http://www.soue.org.uk/souenews/issue4/wallis.html

22

You might also like