back   back to menu

English english language

2.4 The Old Bridge original design

Undoubtedly a detailed historical and typological knowledge of an ancient construction like the Old Bridge and its towers is an important step before any remedial work; this also because a complete understanding of the original design and of the following developing that the construction underwent are precious for any restoration work. Moreover it is possible that the Old Bridge of Mostar was erected following a detailed plan characterised by a modularity with ancient unit measure of the time and by special proportions or relationships among its geometrical forms. This, for sure, was an outspread methodology for composition in the ancient times and it usually may be determined by geometrical analysis.

But unfortunately the Bridge of Mostar seems to be a more difficult case since:

  • the original design drawings of the time are not available (as long as we know)
  • the bridge may have also been changed during its long period of construction (from the original design or plans)
  • the bridge underwent to settlements and form changes during its life
  • the bridge doesn't exist any more and this inquiries are therefore more difficult to be held

To the above listed matters it should be added that there are no historical notes, (as far as we know), that may help this analysis work. Nevertheless, about this subject, many hypothesis may be worked out and partially verified, and by the time other findings come to our knowledge this issue may be updated and mastered.

Anyway, in the aim of making things of easier evaluation, the analysis may be subdivided in three different parts: one is about the original curve of the intrados (shape), another is about modularity and proportions and the last one could be about the symmetrical and asymmetrical elements of the bridge.


2.4.1 Original curve of the bridge intrados

The basic data from which we may begin our inquiry are the X-Y co-ordinates coming form the geometry determination work, either the ones of the north side, either the ones of the south side. Trying to understand which was the original shape of that curve means simply to compare those points with other known curve equations. Of course, this comparing work should consider that:

  • easy and simple results are the most likely to be the real ones
  • at the time they generally referred at a few and simple curves
  • that the curve we are looking for has been probably changed due to settlements
  • that changes may have also taken places also to suit the design to the site requirements

All these variables compel us to express only hypothesis about the original curve, and what it can be said is that, following the indications given by Prof.arch. Carlo Blasi in the book titled "Mostar: Urban Heritage map and Rehabilitation Plan of Stari Grad", the shape of a circle (with its center lowered compared to the bridge springers) gives good results and fits most of the points that may have not shifted.

What it is almost proved is that the shape of the bridge before its destruction had many irregularities, and from further analysis, held in the geometry determination work, it has been come to knowledge that different groups of voussoirs were oriented towards different points due to the above mentioned deformations of the intrados. This matter has been also underlined (following graphical and geometrical methods) by Prof.Dr. Milan Gojkovic that at the same time has confirmed the difficulties in determining the original shape: "However, in the case of the Old Bridge, it is not easy to identify the vault's form after so many years of its existence".

The main conclusion that may be gathered after this preliminary research work is that: in the case that originally the bridge was conceived as a refine and calibrated composition of different curves with centers very close to each other, this will not easily be demonstrated with the available data, and that other simpler hypothesis like the one of the circle are at the moment reliable but not completely satisfactory.

Additional studies, anyhow, have been performed about this subject and have led to other interesting hypothesis which have been numerically documented and without which the analysis of the bridge original design wouldn’t have been carried in the proper way.

general data

span (cm)

raise (cm)

east impost level

west impost level

north elevation



46.71 (m)

46.84 (m)

south elevation



46.72 (m)

46.84 (m)

Throughout a software routine, the same used and described in 5.9 of this report, all the co-ordinate data of the intrados points have been processed in order to find out the circle which could best-fit the intrados points. The above mentioned process has given the following results:

intrados curvature

ray (cm)

center (X) (cm)

center (Y) (cm)

X west shift (cm)

north intrados curvature





south intrados curvature





The above co-ordinates of the centers are related to the relative origins used for the elevations (north elevation = east arch impost; south elevation = east arch impost.

Graphically speaking the result that has been obtained, trough a superposition of the elevations with the related circles, have given the following results:

  • circle centers are lowered of about cm 294-298 compared to the arch impost line level
  • circle centers are shifted towards the west side of cm 7.5-10 in respect of the span axe
  • In both cases (north and south) circles match fairly well the intrados points located in two wide sectors (sector A: about 58) which start from about 18 to 76;
  • Sectors B and C of the circles are not perfectly matching the arch intrados curve: sectors B are higher (external to the circle), while sectors C are inside the circle’s profile.
  • The lowering of the center is of a defined quantity which is linked to the ray by a ratio: lowering distance is equal to 1/5 of the circle’s ray: (north side: 1489/5=297.8 south side: 1477/5=295.4)

fig_21.gif (4038 bytes)

fig.21 – comparison between north elevation intrados curve and best fit circle

The above mentioned observations lead us to think that other curves with other centers have been used for sectors C and B, and a geometrical construction may have used for this purpose.

Of course same results may be obtained throughout the use of similar geometric constructions and what it is going to be explained may be considered only as an hypothesis on the matter, and, in future, additional studies may improve this theoretical approach.

It seems most likely that the others circle's centers were located on the arch impost level since their curvatures are much more accentuated in respect of the main one found, and their rays look smaller. For what concern C sectors the curvature rays are close to a measure which is in the range of cm580-590 depending on north and south elevations. This dimension is close to respect the following ratio: sectors C rays = (5/2) sectors A rays

we have in fact: (north side: 1489/2.5=595.6 south side: 1477/2.5=590.8)

For what concern sectors B, instead, we have that a good matching with the arch intrados curve is obtained with a circle with a center located on the axe at the impost level with a ray equal to the arch raise. The simple geometrical construction is here next represented with an hypothesis on the method that might have been followed in order to determine the limits of the different sectors and curvatures.

fig_22.gif (12231 bytes)

fig.22 – geometrical construction that may have been used for the drawing of the bridge curvature: A,B,C, are the centers of the circles that define the related sectors of previous image. Arc connecting C,A,C may be a construction curve used to determine the sectors limits (intersection with impost level). North elevation has been used for the above example but similar results are obtained for south intrados curvature.

This construction also explains the commonly spread out impression that wants the bridge arch as a peak arch and not as a round arch: sector B is really higher than the round circle mathematically found as the best fit of all the intrados co-ordinates. Of course we can not be sure that this was a real will of the designer, and we can not, as well be sure that sector B was actually composed of only one circle with no peak at the key stone. In fact while north elevation appeared to be flat and continuous, on south elevation there is an asymmetrical irregularity at the key stone level that make us suspect of a peak. In that case sector B would have been defined by two twin circles with their centers quite close one to each other (mirrored to the construction axe). Refer to paragraph 2.3.3 for more details about the bridge key stone point.

Another useful contribute about the examined subject has been given by the preliminary study of Prof.dr. Milan Gojkovic (ICE member). At the time, (February 1999), only preliminary survey were available, but anyhow prof. Gojkovic has come to interesting results with a similar study that hypothesises an higher number of centers. Of course the more centers we use the more we will be able to draw an arch curve which can matches the original one; nevertheless we can not absolutely exclude that the architect, Hayruddin, used a quite complex geometrical construction for the bridge, (even if in that case it would be much more difficult to determine it univocally).

fig.23 – A study of modularity and bridge curvature held by Prof.dr. Milan Gojkovic (Belgrade February 1999) - In background, as a reference survey, a photogrammetric processing by General Engineering.

Anyhow, in recent times, before destruction of the bridge, the curvature was affected by many irregularities, as it has been already pointed out, and all the aforesaid observations should be considered only as a study of the original shape which has nothing to share with the current design which is aimed at the reconstruction of the bridge as it has been documented trough recent surveys (during the XX century). Of course geometrical observations on the curvatures coming from those surveys are quite different and shouldn’t be related to the objectives of current paragraph, (refer to paragraphs 5.6.9 and 5.9 of this report for more notes about recent intrados curvatures). The bridge curvature, as it has come to our time, may lead us to wrong interpretations of what was the original design, which should be gathered from those data trough a global purge of details and trough an extrapolation of the most meaningful geometrical aspects that may be located on both elevations.

To better understand what was the original design of Hayruddin we should ask to ourselves which were the reasons that at the time compelled the architect to perform such choices: why did he lowered the arch center? why did he use more than a center for the curvature? If we are able to give answer to the above questions we may easily gathered what was the original design and what were only construction irregularities or settlements. As far as we know, the bridge design seems more determined by functional devices than esthetical principles and proportions, (see also next paragraphs), therefore it should be possible to find a technical reason for each choice.

Following a strictly functional and practical approach Hayruddin should have used a round arch of a ray equal to half of the span, but in that case he would have reached, due to the arch raising, a very high level at the key stone (fig.24a) that was not suitable for the adjacent streets. By lowering the whole round arch, the pedestrian path levels would have been suitable with the arch raising, but the impost would have been too low compared with the river water that often gets high (fig.24b).

fig_24.gif (10110 bytes)

fig.24 - scheme a: round arch starting from the same springers - scheme b: round arch with lowered springers

Therefore Hayruddin may have found of great use lowering the arch center optimising, this way, either the path level, either the impost level of the bridge.

But he still had a problem to fix: he had to find the way to connect properly the arch intrados shape, that was going faraway inside, with the straight profile of the abutment wall (fig.25a). The use of a curvature with two additional centres allowed him to turn suddenly the bridge profile inwards, and the bridge looked as if it were springing out from the impost level even with its main centre lowered of about 3 meters (fig.25b). This may be easily observed even nowadays: at the springer level the first row of voussoirs, (on both banks), are straight and vertical as it happen in tangency point for a curvature: the center of the circle in that area is therefore located at the arch springer level.

For what concern the top of the arch, with the use of additional curvature(s), no practical explanation has been found, and in this case, the architect may have looked for a delicate, slightly correction of the profile in order to give to the bridge a peak and a raising or a sign which would have reminded some typical non-structural Ottoman arch shapes.

In conclusion it may be said that the bridge curvature original design is probably mostly determined by specific needs and technical devices which were aimed at inserting suitably the structure in the natural site where two constraints were unmoveable for morphological reasons: the span and the adjacent streets levels.

fig_25.gif (9890 bytes)

fig.25 - scheme a: arch defined by a curve with the center lowered - scheme b: additional curvatures at the impost level.


2.4.2 Modularity, proportions and geometrical relationships

If any modularity was present in the bridge of Mostar, it can be investigated, first of all, trough a dimension analysis: in other words if there is a dimension that is repeated many times for different elements, or it is repeated divided or multiplied per whole numbers this could be interpreted as a dimension modularity. Unfortunately the bridge is composed by few typological elements and few groups of dimensions: like voussoirs, ashlars, parapets, and global measures like intrados thickness, span and distance from key stone to springers. There are not many significant groups of measures that may be successfully compared looking for relationships; the only thing that may be underlined is that the measure of about cm 78-82 is quite frequent: it is present in the arch voussoirs as a whole and as half of it (average dimensions cm 8040), in the intrados thickness (5 times - cm 400) and in the bridge span (36 times). But this actually doesn't prove much.

The analysis may be better held comparing the unit measures of the time with the surveyed dimensions, but even this doesn't seem an easy task: at the time there was not a declared unit measure and there were changes due to the places and due to the periods. As long as it concerns the Old Bridge of Mostar, for instance, we have from researches of Prof.dr. Milan Gojkovic: arsin of about cm 62-63 (contained 46 times in the bridge span); from researches of Prof.arch. Carlo Blasi: arsin of about cm 71-75 (contained 40 times in the bridge span).

Both of the above may be reliable hypothesis, even if the bridge span could not be the most meaningful distance to which make comparisons: it could have been influenced or modified to fit the distance coming from the morphology of the site.

Geometrical relationships investigated by comparing the shape of the bridge with elementary figures like triangles, squares and circles are possible and it is reliable that the original design followed this sort of geometrical rules. But being everything (about the original geometry) still so much uncertain it may be premature to work out any relationship of the kind: the matter may be better analysed once more findings come to our knowledge.


2.4.3 Symmetrical and asymmetrical elements of the bridge

Talking about the original shape or original design, it has to be noted that the bridge, before its destruction, was mainly a symmetrical structure apart from some small irregularities and deformations due to settlements. But by the west bank it is possible to note that a portion of masonry of the elevations has got an oblique layout as if some ashlars were added in a second time. This reveals a possible different level of the pedestrian path and a consequent different arrival level of the bridge in the first configuration of the construction. So we are brought to think that maybe the bridge originally was not conceived as a completely symmetrical structure (also the stiffening ribs of the load bearing vault had different sections, as it can be noticed now that the bridge has ruined). And it is probable that the bridge on one side was so steeply (about 27%) that it was difficult to cross it with animals and wheelbarrows.

If this peculiarity would be confirmed also by the archaeological findings, then we may start thinking that the original design was mostly influenced by the natural morphological characteristics of the site, by practical requirements and not only by proportions, modularity and esthetical values.

fig_26.gif (15109 bytes)

fig.26 – Hypothesis on a possible original configuration of the bridge compared to the recent known shape of it

Another element that leads to an asymmetrical configuration of the bridge is related to the impost levels: the arch springer by the east side is lowered of about 12-13 centimetres in respect of the west one. The above mentioned irregularity seems to be too high to be considered a settlement of the rocks of the bank, and it is most likely that wholly, or in part, was caused by ordinary construction inaccuracies, also because the bridge structure has been assembled with high care for what concern technological and strengthening devices but regardless of some other global alignments.


2.5 Hypothesis about the ancient yard and site

As far as it has come to knowledge, not much is available about the ancient yard and the ancient assembling and constructive techniques of the time, nevertheless it is possible to gather many information trough a detailed research work on the site and on the bridge remnants. The archaeological research work held by Mr. Bessac is undoubtedly a prove that it is possible to develop reliable hypothesis on ancient techniques that are absolutely not documented in any paper. But also the architectural research work and analysis may give its useful contribute to what should be considered one of the main aims of a rehabilitation design.

In this report chapter 9 is mostly related to the ancient assembling techniques: how they were performed, in order to determine how they will have to be performed. Moreover, in the following paragraphs some additional notes are given to contribute towards the description of the ancient constructive technique of building a stone bridge: these notes are mainly the results of numerical observations held on the geometrical data.


2.5.1 Hypothesis about the ancient constructive and assembling techniques

Building a stone bridge, more then 400 years ago, was really a feat and most probably the highest difficulties of the time were the following:

  • structural and architectural design was not determined trough a knowledge of the theoretical principles related to stability and resistance but trough the knowledge of the causes of collapses and effects of loads: therefore anything new would have meant something unforeseeable;
  • the building of a wooden false work, (centering), that could bear the huge load of the stone blocks with limited settlings has always been one of the most remarkable issues of large ancient stone structures;
  • transporting and lifting heavy stone blocks was a technological issue that was strictly linked to the times of performing the works which had to be controlled to avoid collapsing risks related to the season variations of the river water level.

Unlikely, no historical document is yet available about the tale of the bridge yard, with organisation notes, constructive methods, difficulties found: therefore, what above mentioned, is mainly an hypothesis based on similar cases.

But what follows is more scientifically proved by the architectural analysis held on the ancient bridge surveys.

The centering was most probably a wooden one; nothing is known about the exact configuration of it, apart from the fact that some recurrent discontinuities have been found in the arch intrados, which may have been caused by different sectors of the arch false work, (centering - scaffolding); these sectors may have been defined by the length of the master beams used for the structure.

It is most likely that the ancient centering was not strong enough to bear the loads of the stone blocks without settlings: this may be proved by the fact that, by comparing north and south elevation, it is possible to note, (refer also to 7.5.1), that next to the springers the rows were much more regular and well built, with little variations of levels. While, proceeding to the key stones, irregularities increase: rows are proceeding not parallel, levels are changing from north to south in a range of about 10 centimetres. Presumably, the more the centering was loaded, the more it underwent to unforeseen settlings that were more remarkable by the south side, as it can be observed with the numerical analysis of the co-ordinates. To recover the settlings, it is possible that wooden wedges were used, of which there are traces in many spots of the intrados surface of the vault, (some steps between adjacent voussoirs may be surveyed even in the assembled blocks recovered from the river). The use of wooden wedges was probably performed with very small ones next to the springers in order to adjust the voussoirs and to match their joints and was performed with bigger ones towards the top, to correct the geometry following the planned design shape.

Even if arch stones were of remarkably different dimensions, the preparing and assembling procedures were not randomly performed, (refer to chapter 9 for more notes), and what is even more important to stress, the work over the centering was most likely to be quite limited to the assembling: in other words most of the work like stone cut and carvings of slots were performed off-site. The above statement may be proved by the following observations:

  • positioning of voussoir joints was, as in ordinary masonry works, accurately shifted to guarantee an efficient interconnection of the stone blocks;
  • positioning of cramps and dowels, and related slots and channels, required the exact knowledge of the stone dimensions and of their joints positions of adjacent rows of voussoirs to avoid interference between metal strengthening devices and joints;
  • stone voussoirs were quite variable in the vault, but the ones belonging to the same row were of very close transversal base dimensions: average variation range cm 0.5-2 in a length of almost 4 metres, (refer to 7.5.5), which is an accuracy quite higher than the average followed for all the other parameters.

From the above observations it is possible to deduce that, at the time, they were using different stone voussoirs for the bridge vault of different size and shapes, (due to the natural availability of the quarry, where it seems that some natural weakness veins compelled to limit the dimensions), but each row of the vault was accurately pre-selected and picked from a temporary deposit of rough blocks. Each row was composed of voussoirs of very close intrados dimensions, and was prepared next to the preceding one in regard of the joint positioning and of the metal elements positioning. Dowels were previously assembled off-site and related slots were prepared. This procedure was most likely to be performed on groups of rows and not on couples to avoid that the assembling could be stopped by the lack of prepared arch rows. Over the centering it wouldn't have been possible to manage all the vault requirements unless an efficient communication of dimensions were performed between the working teams over the centering and the working team off-site.

Despite the accuracy concerning the transversal thickness of arch rows was accurately performed, the raising of the rows was quite irregular towards the top, (refer to 7.5.4), most probably due to the mentioned centering anomalous settlings. This increasing inaccuracy has been recovered wholly in the three top rows at the key stone level, where assembling has been performed regardless of the previously surveyed criteria, and even a variation of cm 11 of the intrados size has been checked in only one row.

Most probably the arch top rows of voussoirs are the prove that the two teams of workers were not in contact and not co-ordinated in the carrying-on of their works. It may be possible, also, that something was going wrong with the centering and that final rows have been quickly assembled to stop the gradual settling of the vault. We shouldn't forget that the vault was about 145 m3 of stone which weighted almost 300 tons over a wooden temporary structure.

For more notes about this subject refer either to chapter 7, either to chapter 9 of this report.


2.5.2 Hypothesis about the ancient site configuration

With recent findings, studies, research works and archaeological investigations we are quite close to determine a possible ancient configuration of the site and related developments. Still some information are missing and everybody is waiting for historical notes about the site and the towers, plus some additional archaeological researches of the abutment walls.

Nevertheless, at the moment, it seems possible that the Hercegusa tower, together with a symmetrical buttress, round shaped, found in the inner structure of the north-east abutment could have been, together, an ancient fortification castled over the east bank: in the axe of them the Tara tower is located.

Undoubtedly the abutment walls, by the north east side have been developed in subsequent layers to protect the fortification structures from the river thrust and erosion. Behind them it will be possible to find the most ancient configuration of the site, and may be it will be possible to find exactly the location where the ancient bridge, suspended on chains and bequeathed by history, spanned the Neretva river before the Stari Most. Its location may be not far from the spot and may have been in-between the above mentioned fortification buttress.

By the west side the research is even more interesting: from ancient drawings useful notes may be gathered and the configuration of the site seems to have undergone to some remarkable changes; but also in regard to the notes of the 2.4.3 of this report it would be really of use to determine definitely the ancient morphology of the bank and its levels.


Intellectual property of this report and of the design drawings is owned by General Engineering s.r.l.

author of the text: arch. Manfredo Romeo – other contributes have been mentioned in related paragraphs

- General Engineering Workgroup -


Final Design Report

back   back to menu
ani.gif (16635 bytes)
GENERAL ENGINEERING - P.zzale Donatello 4 - 50132 Firenze - Italy - ph. +39 055 2345256 - fax. +39 055 2476074