A thesis on the structural assessment of Guastavino domes cannot omit an overview of Rafael Guastavino, his company and cohesive construction. Accordingly, a brief introduction is provided to familiarize the reader with the cohesive construction system. Subsequently, based on the knowledge gained through this research, questions that arose in the problem statement regarding the existence of horizontal thrust and the elastic behavior of the domes are answered. The discussion continues with the contribution of Guastavino to the architectural design development of the City County Building based on the documentation obtained from the Carnegie Mellon University architecture archives.

6.1 Literature on Guastavino and Cohesive Construction

Prior to initiating a research study on Guastavino domes, a survey of the literature presented by him and other scholars with an interest in his life and work is essential.

6.1.1 Rafael Guastavino

Rafael Guastavino, born in Valencia in the year 1842, emigrated to the United States in the 1880’s, thus importing the technique of erecting thin masonry, a thousand-year-old building system of ‘Catalan vaulting,’ from Spain (Rossell 2002). Although the question of the system’s historical origins remains unanswered, it is important to mention that the timbrel vaulting is referred to by many different names in different regions: in Italy, volta a foglia (layered vaults); in  France, voûte plate (flat vault); in Spain , boveda catalan (Catalan vault) (Lane 2001).

Wherever the origin, the importance of the refinement that this type of vaulting underwent in Spain for the last hundred years is unquestionable for the career of Guastavino (Collins 1968). Guastavino professed that the cohesive construction technique was not a discovery but rather a development: “Neither the brick vaults nor timbrel vaults can be said to belong to any civilization. These circumstances necessitated the creation of them in every country” (Guastavino 1892).

Guastavino claimed to be the first and only one investigating the structural abilities of Catalan vaulting: “In Spain where this system has been used and is still in use on a larger scale than in any other country, there does not exist any treatise, nor a single work on the theory of this construction, nor a single scientific explanation can satisfy curiosity” (Guastavino 1892). He again emphasized this aspect with these words: “…In Spain 95 % of the architects and 99% of the builders did not know or may not have heard anything about this system” (Guastavino 1892).

Guastavino –probably not as alone as he claimed– studied the building system of terracotta and introduced refinements that led to the technique he referred to as “cohesive construction.” His greatest refinement on the system–altering the traditional mortar with rapidly hardening portland cement–enabled him to build vaults that spanned three to five times the typical span of traditional Catalan vaulting (Tarrago 2002). He decided that availability of stronger and more consistent mortar was an essential component for the development of this system, and hence he had continuous trials on mortars throughout his career. Guastavino was an experimenter and a supporter of technological developments, as can be appreciated in his call for assistance to improve the cohesive construction technique (Guastavino 1892).

I must remark that I have yet a great deal to do by way of improvement, and it is necessary that I call for assistance in perfecting our knowledge of the art of building, especially from manufacturers of materials and architects.

In 1824, Portland cement based stucco was patented by Joseph Aspdin. Aspdin’s Portland cement does not require exposure to air for setting, and thus it was able to set between two tile units. Guastavino’s writings claim that this development in cement production commenced the renaissance of the “cohesive construction.”  He eventually opened a production plant and manufactured the tile and mortar to be used in his construction. There is a long-standing belief that Guastavino used some kind of special additive in his productions or had a secret recipe that he guarded throughout his company’s life. In 1889, the Engineering News published an article on Guastavino and used these words to describe his mortar (Engineering News 1889):

…the first layer of tiles is laid with a quick-setting mortar, composed principally of plaster-of-Paris. The other layers are laid in mortar composed chiefly of Portland cement. Its exact composition is a secret; but it adheres so closely to the tile and is so firm and solid when it has fully hardened that its strength is about equal to that of the tile itself, and the whole arch is practically a monolithic mass, showing no tendency to separate at the joints more than at any other point.

Guastavino argued passionately that the masonry construction system is superior to any other, and the revival of masonry architecture would satisfy the requirements of the present day and be the main construction material in the future. With the help of previously built examples, Guastavino promoted cohesive construction in North America and attained a strong reputation in the field of cohesive construction techniques, not only as a contractor, but also as a consultant (Wright 1901).

6.1.2 Cohesive Construction

The cohesive construction system is best understood when compared to conventional masonry vaulting [Figure 6-2]. Guastavino vaults are composed of multiple layers of 1” thick tiles (usually 6”x10” or 8”x12”) laid flat in 1” thick Portland cement, instead of a single layer of vertically oriented wedge-shaped thick cut stones. The terracotta tiles are bonded to each other by the tensile strength of cement beds, while the stone voussoirs hold their integrity by compression and friction between elements.

Guastavino presented a paper in which he professed that in cohesive construction the tiles are bonded to each other with cohesive forces due to the assimilation of tile and Portland cement. According to his statement, multiple tile layers are necessary to break the joints in order to invoke the cohesive forces, (Guastavino 1892):

A ‘timbrel vault’ of a single thickness of tile has no more resistance than an arch or vault built in the ‘Gravity System’, because no matter how good the mortar may be there is only one vertical joint, and the tiles are working as voussoirs…

…But if we put another course over the first one breaking joints and laid with hydraulic material, we will have the action of the cohesive force: in this way the mortar laid over the first course, or extrados, takes bond with it and also with the 146  course laid on top. As soon as the mortar is set, we have shearing resistance represented by 17,820 pounds per square foot.

In the cohesive system, the tiles are typically arranged so that they are staggered with respect to the joints of the previous layer. It also is a common practice to shift the joints 45 degrees at each layer to increase the strength of the vaults [Figure 6-2].

One of the most remarkable aspects of Guastavino vaults is that they hold their own weight the day after the tiles are set. Due to this inherent self-supporting nature of the construction, the vaults can be built without centering and scaffolding – unless it is necessary to elevate the workers. The process of building a Guastavino dome can be enumerated as follows (Etheredge 1971) [Figure 6-3]:

• The first course is laid skewback on a groove in the arch using plaster of Paris. The worker holds the tile until this fast setting mortar on two edges holds the tile unit. At this stage, a wooden template of one tile width is usually employed to assure the proper curvature.

• Once the first course is laid, the second course of single tiles is progressedbreaking the joints of the previous layer with rich Portland cement mortar.

• After the second course is set, the worker can stand on the tile ring andlean to progress erecting the first course of the second row, which is again built with plaster of Paris.

• The procedure is followed until the dome is closed at the apex.
• Once the first two layers are completed, it serves as a formwork for the subsequent layers of tile and Portland cement. Thus, in cohesive construction, no formwork or centering is necessary; indeed centering is intentionally avoided since wood, by absorbing the moisture content of mortar, expands and degrades the curvature. Moreover, the expansion of the wooden formwork also decreases the compressive forces desired for the setting of the mortar.

During the decades of Guastavino’s practice in the American construction market, incomplete tile vaults carrying their own weight was noteworthy. However, the idea that the few inch-thick domes could bear significant loads was not easily accepted by the public (Lane 2000). Therefore, Guastavino, in order to establish confidence in his system, needed to undertake some experiments.

Guastavino conducted tests in order to prove the performance of this system to the public [Figure 6-4]. In Manhattan (1901) the 10 feet span Guastavino vault of four courses of 1” thick tiles sustained a 563-ton load in a load test (Collins 1968).

Guastavino indicates a series of tests conducted with engineer A. V. Abbott at Fairbanks Scale Company in May 1887 (Guastavino 1892). Although no detailed information is provided regarding the test setup or methodology, obtained values for four different sets of compression tests are presented [Table 6-1]. The significantly varying results are then linearly averaged as 2,060 psi (14 MPa). Additional test results on shear, tension, and transverse shear are also presented without further detail. The difficulty in sustaining stable properties for these products even in laboratory conditions is apparent in the test results. Guastavino subsequently used the values obtained from these tests as the design coefficients in his future work [Table 6-2]. 

Table 6-1: The compression test results obtained at Fairbanks Scale Company in May 1887 (Guastavino 1892).

Prof. Gaetano Lanza, PhD., professor of applied mechanics at M.I.T, prepared a table of stresses occurring in a 10% rise arch under uniform unity loading [Table 6-3]. There is no additional information on the calculations of Lanza available in literature.

Table 6-2: The complete test results obtained at Fairbanks Scale Company in May 1887.

Guastavino’s writings do not provide clear information on the production of either tile or mortar. However, in a few instances he briefly discusses some specific issues in cohesive construction. This attitude might be due to the changing nature of the materials in those days or that perhaps he did not want to share the full recipes with his rivals.

Table 6-3: Theoretical stresses for arches of 10% rise with uniform distributed load per
ft² by Prof. Lanza (Guastavino, 1892).

In his first essay, Guastavino explains the main topics that need to be improved to perfect cohesive construction, one of which was the tile manufacturing. He states that assuring strength of 2000 psi, the tiles manufactured for cohesive construction need to be as light as possible so that they can float in water (Guastavino 1904):

To the manufacturers: our tiles as well as our bricks are too heavy…it is an error to believe that the heavy brick is the best; the brick which I mentioned, has a breaking attain of 2000 pounds and will float in water…

Among other things, Guastavino states that the tile and mortar assembly has a robust nature for construction irregularities (Guastavino 1904). Doubling the amount of water for a mortar lump would halve its strength due to the fact that evaporation of the excess water results in voids within the cement. However, if tile and mortar are put together, in Guastavino’s words, “the clay absorbs the water in excess, leaving the moisture necessary for the requirements of setting the cement.” He also assigns certain strength criteria for mortar:

The joints have great importance in brick masonry.  The mortar used must have as great strength to resist pressure as the brick.  If the brick after being burnt has a resistance of 2000 psi, crushing load, the mortar used should be of equal resistance after ‘thirty days’ use (Guastavino 1904).

Although Guastavino obtained the coefficients from the tests, his work was still empirical since there were no tools to deliver the definite calculations of his engineering. Guastavino wrote in his first book that; “The thickness of the arches was determined by intuition and practice…” (Guastavino 1892). In time the Guastavino Company, perhaps based on their experience, developed empirical standards on the number of courses necessary for arches and domes under distributed loading [Table 6-4] (Engineering News 1889). Even though the standards might show potential for today’s engineers analyzing Guastavino domes, caution is necessary while adopting them since Guastavino is inconsistent about the number of courses necessary for a dome in different sources.

A careful look at Lanza’s table illustrates significantly different information on the tile thickness for a Guastavino arch [Table 6-3]. Lanza’s table, published in 1892, displays exactly the same values for the same span ranges as in the article in Engineering News, published in 1889.  However, Lanza’s table is in inch units, not in the number of courses. Despite the resemblance in the numeric values for the thickness, these two charts do not result in the similar “safe load” solutions. This discrepancy can be a simple publication mistake, a result of the learning process, or again it may even be intentional to misinform and confuse the company rivals.

Table 6-4: The Guastavino Fireproof Construction Company’s specifications for erecting tile arches and domes, Engineering News, November 9, 1889.

Although a misprinting error in a weekly journal seems to be the easiest explanation, this does not resolve all the anomalies in Lanza’s table. As both tile and mortar are 1” thick in this construction system, it is surprising to see thickness values starting from 2” and including the subsequent even numbers: 4” and 6” [Table 6-3].

Lanza’s work is limited to Guastavino arches; however, the charts in Engineering News include standards for the domes as well.  According to the article, the standard numbers of courses for the domes are one less than those of arches. Once more, we come across contradictory information in the 1892 article of Guastavino. He states; “…it is sufficient to have only two courses of tiles, one inch thick, for dome of sixteen to twenty feet span, introducing some cases ribs for extra strength” (Guastavino 1892). However, the standards in Engineering News require five courses of tile for a dome of sixteen to twenty feet span.

The greatest span ever achieved in the company’s life is the crossing dome of the Cathedral of St. John the Divine in New York City. At 98 feet in diameter, this dome demonstrates superior engineering skills. The construction drawings of the Guastavino Company illustrate reinforcement around the abutments. The dome is built with a maximum thickness of 12”, which is less than the required thickness for 20'-24' diameter domes presented as standards of the company.

Like so much of Guastavino’s writings, it is not clear where these values are calculated on the dome. Remembering that it was common to lessen the course numbers towards the crown, one can only speculate that these two different charts were calculated for different locations on the dome.

Another peculiar characteristic of Guastavino vaulting is that, unlike conventional stone construction, the components cannot be separated without destroying the integral vault shell. As Guastavino puts it; “the structures built by the ‘Gravity System’ can be at any time taken down, in the pieces out of which they were formed….while man cannot again use the parts of cohesive construction.”  Collins observed the removal of the vaults of New York’s Metropolitan Museum due to the insufficient quality of the tile vaulting completed by contractors other than Guastavino. Collins stated that not a single whole tile unit existed in the demolition site (Collins 1968) [Figure 6-5].

A further distinctive advantage of the Guastavino vaulting system is that it can be perforated without losing its structural integrity, as seen through an accident during the construction of the Boston Public Library in 1892 [Figure 6-6]. Two tons of weight fell and punched through a section of the vault and rib. The structure survived due to the inherent redundancy even though it lost a large portion of its mass.

As the tiles protect the hydraulic mortar from excessive heat, the cohesive construction system is fire resistant. Guastavino, a very good marketer, certainly profited from the concern over fireproofing and building code requirements of the early 19th century, which came about due to the severe fires in Boston and Chicago. Consequently, he changed the name of his company to “Guastavino Fireproof Construction Company.”

The company was involved in the construction of more than 1000 buildings in 41 states during the lifetimes of two Rafael Guastavinos; the father (who will be hereafter designated as Guastavino I) and the son. Many of these buildings are found in New York, Boston and Pittsburgh (Collins 1968). Figure 6-7 presents an advertisement of the company including some of the domes constructed with this system. During their period of activity, the two Guastavinos seem to have maintained an absolute monopoly on tile vaulting in North America, possibly due to their numerous patents on construction techniques. Guastavino’s brand of vaulting disappeared shortly after the death of Guastavino II in 1943, due in equal probability to the monopoly of the Guastavino Company in the market, the secret recipes of mortar and the increasing cost of hand labor.

6.2 Discussion of Guastavino System

The following sections present a discussion of the Guastavino vaulting system in the context of structural engineering. The statements are done based on the results obtained through the present research and knowledge gained through literature review.

6.2.1 Presence of Horizontal Thrust

Guastavino and other scholars are of the opinion that a cohesive dome does not exert any horizontal thrust on its supports due to its monolithic nature. This argument is repeated in many published papers with phrases like “no-thrust’ and ‘minimal thrust’ (Collins 1968, Neumann 1999). Guastavino states this claim in his first essay (Guastavino 1892):

Suppose we take a big block of stone, say then feet long and ten feet wide, and one foot or one foot 6 inches thick; if we support that on the four sides just as a lintel we have practically no thrust, and if we make a cavity on the under side, making a curve like dome, we will have a dome arch but will have no thrust.

Interestingly, again in his first essay, he corrects the misunderstanding on the absence of horizontal thrust for tile arches (Guastavino 1892):

It is frequently seen that the greatest friends of the system sometimes go too far in their enthusiasm and favor of the new idea…for instance it is said ‘that arches under system have no thrust.

...the barrel arch has some thrust, and requires some material to counterbalance this, that is rods. That is one of the causes, which makes the barrel construction more expensive than the dome...

If the cohesive domes were perfectly rigid, as in the analogy of the big block of stone, Guastavinos argument would be valid; however, the analogous stone block actually needs to be carved on both the bottom and the top to be applicable to cohesive domes. Noting that the cohesive domes are far from being perfectly rigid, one can easily see the fallacy in the argument. The force distribution in the stone block is reoriented when it is carved, resulting in a horizontal component at the periphery.

It is interesting to note that, despite his passionate claims, horizontal thrust is in fact taken into account in Guastavino’s calculations and erection of domes, as presented in Figure 6-8. His formulation for cohesive arches is based on the simple statics of the voussoir arch [Equation 6-1]. The gravity load of the voussoirs generates pressure against one another, thus arches and vaults built in the gravity system exert a lateral thrust that must be counteracted by the walls or buttresses.

(Equation 6-1)

His equations for domes are extensions of the equations for arches, assuming that one-half of the loads are carried in each direction, longitudinal and transverse [Equation 6-2]. Therefore, Guastavino calculates the center thickness of a dome with the formula for the arch and divides by two. In all cases, the existence of the lateral thrust at the support is considered in his equations.

 (Equation 6-2)

The results revealed by the ANSYS model solutions confirm the existence of a large horizontal thrust exerted to the adjacent members by the dome. In SEB, the vertical load transferred to one pier is almost equal to the horizontal thrust exerted to the adjacent dome. For CCB, the tile arches support a horizontal load, which is equal to 75% of the vertical load transferred to one pier. The values obtained through this study point out the fact that the horizontal thrust is a significant component in Guastavino domes and need to be counterbalanced. Practicing engineers who are assigned preservation work on Guastavino domes need to take the horizontal thrust into account.

Lane (2000), in his M.S. thesis on historic preservation, discusses the conflicting arguments on the lateral thrust of Guastavino domes. He presents the opinions of Luis Moya Blanco and Bassegoda Musté on the issue. Without any investigation based on the theories of structural engineering, he concludes that cohesive vaults exert a thrust but less so than voussoir vaults due to lighter mass. Lane’s statements also go further; he suggests that Guastavino, knowing the truth, took advantage of the market while his system was new. Again, no solid proof for this argument is provided. The next section will show that the two Guastavinos were well aware of the existence of the lateral thrust in dome structures.

6.2.2 Reinforcement of Tile Domes

The fact that monolithic shell domes with tensile resistance always have a lateral thrust is not a remarkable discovery in structural engineering. However, present study provides a solid base for a discussion on Guastavino’s attitude, important for the disciplines of art history and historic preservation.  In this section, the two Guastavinos –father and son– are evaluated separately.

What is noteworthy in Guastavino I’s essays is inconsistency. For instance, his writings deliver conflicting information on the standard (or suggested) tile thicknesses for domes within different articles or applications. It is not clear whether this inconsistency is representative of a learning process or a marketing plan used to promote the system in any possible way.

Although Guastavino I confidently argued that, his domes do not have thrust due to the monolithic nature of the tile and mortar assembly, Guastavino II apparently did not put all his faith into the tensile capacity of the cohesive system.

In July 31, 1908, four months after Guastavino I’s death, Guastavino II filed an application for the patent on reinforcing the tile arches and domes by steel bars. The steel reinforcement clearly is presented to account for the possibility of tension zones occurring in the dome webbing.

Reinforcing or strengthening rods are embedded in the binding material in the joints between the tiles, so that the bonded strength of the whole structure is not in any way lessened by the introduction of these rods. The rods are preferably of a spiral form, to secure an interlocking engagement with the binding material, and are adapted to extend entirely around the dome and are of such dimensions in cross section that they can be readily introduced between the layers of tile without disturbing the general method or scheme of laying and bonding the tiles to each other (United States Patent Office, 1910).

Perhaps Guastavino II was not very confident about the tensile capacity of tile and mortar assembly, or he wanted to further improve the system to span larger distances with thinner shells. In either case, the patent specification of the reinforcement permits an interesting inference from the following sentence:

As the greatest outward thrust of a semicircular dome is near the base, [emphasis added] I place the metal rod extending around the dome closer near the bottom thereof, as shown in figure 1.

From this sentence, we understand that in 1908 Guastavino II knew of the existence of a lateral thrust in his domes. However, the approach of Guastavino I to the subject is still not clear. It is equally probable that Guastavino I knew the truth his whole life or never understood the behavior of masonry domes. However, remembering that he corrected the common misunderstanding of the public on the cohesive arches, gives credit to the latter–his misinterpretation of the dome behavior and consequently his honesty. If he only wanted to promote his system, he would easily have led the public to believe in the fallacy that cohesive arches do not exert thrust.

On January 18, 1910, Guastavino II was granted the patent on reinforced masonry domes. Guastavino II was clearly well aware of the fallacy in his father’s argument on the “carved stone” analogy. We see the application of this improved system to many later buildings by Guastavino II. The crossing vault of the St. John the Divine, built in 1909, has considerable amount of reinforcement around abutments to account for the tensile stresses.

One would expect to see writings of Guastavino II to correct the misinterpretation of his father and to inform the public on the issue.  Perhaps the possibility of damaging the fame of the company prevented him from making such a public announcement.

6.2.3  ‘Elastic Thin Shell’ Behavior

Although Guastavino I particularly argued against the reinforced concrete system several times in his essays and claimed it to be inferior compared to his system, another well accepted argument on cohesive construction is its resemblance to concrete shell structures. In George Collin’s words, “a sort of concrete made with aggregate of highly regular pieces–the tiles” (Collins 1968).

In Guastavino vaults, the contribution of the Portland cement to the shell volume is 50% or more of the total volume of masonry. The structural integrity of this type of construction depends upon the assimilation of the tile and mortar as a result of a chemical reaction. Guastavino declared that once the mortar between multiple layers of tile sets, the vault shell takes on a monolithic state.

Musté (Musté 1997) agrees with Guastavino on this issue and states that the multiple layers of tile bonded with a layer of mortar leads the assembly into an elastic behavior (translated by Lane, 2001):

The characteristic resistance of these structures manifested itself at the moment in which the shell is doubled via a new layer of tile being added to the first with a good mortar. Only then, does the structure pass categorically into the notable realm of superior biresistance [emphasis added] and continues to become stronger in direct relation to the quality of materials used.  Prof. Cardellach posited that it is not unreasonable to think that in near future the forces of shear and compression of the vaults could be considered equal and they could then achieve the docile flexibility of a sheet of steel.

Although reinforced concrete as a building material was available in Europe since the mid 19th century, during the days of Guastavino’s career, reinforced concrete was still in an experimentation process in North America. The initiation of the revival and the rapid development of reinforced concrete in this continent dates to Ransome’s Leland Stanford Museum in San Francisco in the 1890’s.  It was not until 1907 that the earliest textbook on the theories of reinforced concrete was available. The studies on the practical mathematical means of analyzing doubly curved thin shell structures were not initiated before the development of reinforced concrete shells around the 1920s. Thus, the theories of reinforced concrete or practical applications of thin elastic shell were not available for Guastavino prior to his career.

Although Guastavino was not furnished with the advanced theories of material science or doubly curved shell behavior, his domes, which were designed and executed based on intuition and experience, display a linearly elastic shell behavior provided that they are free from cracks. The present research displayed that the axisymmetric natural frequencies of the domes can be approximated with the vibration formula delivered by the thin elastic shell theory. It is also observed that the nonlinearity introduced by the cracks has a degrading effect. Therefore, as long as the domes are designed appropriately and do not have severe cracks, the theories of thin elastic shell result in reasonable accuracy.

The nature of the masonry as a material requires many approximations, which also causes discrepancy during the comparison of results.  Masonry, when tested with accurate modern techniques of experimental stress analysis, does indeed exhibit inelastic material behavior. This is mostly caused by the permanent damage or deformation after certain amplitude of loading or after a certain amount of cyclic loading. Masonry in highly precise measurements also displays a nonlinear behavior, in other words a non- proportional relationship between force and displacement.

However, in the accuracy range of this study, as stated prior to the discussion of the research, masonry is assumed to remain in linearly elastic range as long as the stress does not exceed the allowable tensile or compressive stress capacity of the material. The static results of the validated ANSYS model displays that the SEB is subject to very low stresses compared to the capacity of tile and mortar assembly, while the CCB has critical stress concentrations, which exceed the allowable stresses given by Guastavino (Guastavino 1892). Additionally the experimental results confirm the linear behavior of the dome of SEB in an acceptable accuracy, while the effects of nonlinearity are evident in the experimental data of the CCB.

As a result, provided that the structure is crack free, the linearly elastic behavior assumption is valid for Guastavino vaults and can subsequently be utilized for further studies. If the masonry voussoir were used instead of the tile and mortar assembly in the SEB and the CCB, the mortar joints of zero tensile capacity may yield hinges at certain tensile stress locations, and the behavior would be more likely to be nonlinear and inelastic. Therefore, Guastavino vaulting is conceptually closer to a concrete membrane than the gravity system.

6.2.4 Design Interaction of Guastavino and Hornbostel

The effort to illuminate the design interaction between Guastavino and Hornbostel and his associate architects is limited to the City County Building, due to the fact that the immediately available documentation on State Education Building is limited.

The City County Building architectural design was commissioned to Edward B. Lee associated with Palmer, Hornbostel and Jones, architects, through a 16-entry competition in 1914. Although Lee was assigned as the principal architect of the project, he stated in 1948 that the building was executed according to Hornbostel’s design (Kidney 2002). Based on this information, the CCB will be attributed to Hornbostel in this study.

In December 1915, the Allegheny County officers posted an advertisement in the newspaper and invited contactors to submit bids for the timbrel arches. Although the Carnegie Mellon University archive has extensive documentation regarding the contracts and specification of the CCB, Guastavino is the sole bid presented for the timbrel arches. In all probability, Guastavino was the only bidder, or his bid was the only one preserved as he was awarded the contract.

The fourth floor plan drawing approved by the art commission displays an entirely different vaulting type for the Grant Street loggia. As seen in Figure 6-9, the original design consideration for the loggia is not a dome, but a main barrel vault with intersecting smaller vaults (quasi groin vault). The drawings also appear to indicate the joists for a voussoir vault. It is not known at what stage these design alterations occurred, but the construction drawings of the contract construction engineer, James L. Stuart, present a tile dome without buttresses for the vestibule loggia. These drawings also display the volume between the dome and the balcony floor to be filled around the apex [Figure 6-10].

Despite the architectural and construction drawings, the dome built by Guastavino is a tile dome with buttresses and without any filling [Figure 6-11]. It is evident that Guastavino had an influence on the structural design of the domes, but the extent to which he participated in the architectural decisions is not clear.

6.3 Concluding Remarks

Through this research, the tools of modern engineering demonstrate the significance of the horizontal thrust in Guastavino domes. After establishing this fact, the awareness of the Guastavinos on the issue is investigated through several sources. It is found that Guastavino II was knowledgeable on the topic and improved the system by adding steel reinforcements. This improvement was either transferred from father to son or developed solely by Guastavino II. Guastavino I’s attitude in his “no horizontal thrust” claims is still not clear at this point. He either wanted to promote the system in public or did not comprehend the dome behavior.

Guastavino thin shell domes without cracks exhibit linearly elastic behavior, whereas existing cracks engender nonlinear and inelastic behavior. Future researchers should take note that a simple constitutive law may not be sufficient in accuracy for domes with cracks.

Available documentation at the time of publication on the CCB is investigated to understand the design interaction between Hornbostel and Guastavino. It is seen that not only the original architectural design of the loggia domes, but also the construction drawings of the engineer were altered. This fact gives credit to the hypothesis that Guastavino was consulted during the design of the domes, and his comments were influential in the final construction.