**3. Results**

### *3.1. Analyses of Roof Oculi Configuration in the Representative Sample of 13 Heritage Hammams*

During the field work survey, it became clear that all the visited hammams had their roof oculi poorly repaired or completely blocked and none of the original blown glass caps covering the oculi had survived in any of the hammams. Furthermore, the identification from the roof spaces of the original oculi, their number and position was not always possible. This was due to inadequate roof maintenance practices consisting of different layers of cement applied to the roof surfaces. This not only reduced the roof openings' diameter (and the light penetrating to bathing spaces) but completely covered some of the original openings (Figure 5) leading to the use of poor indoor electric lighting during daytime hours. The examination of the hammam ceiling from the inside of the bathing spaces was therefore necessary (although not always easy) to locate and count the blocked oculi.

*Sustainability* **2018**, *10*, 3984

The examination of the architecture of the 13 documented hammams reveals a consistent linear organisation of three successive rectangular parallel bathing spaces of increasing temperature and steam and are as follows:


**Figure 5.** *Cont*.

**Figure 5.** Hammam Sidi Ayyub (16thC) from left to right (**a**) view of the roof; (**b**) internal view of the blocked oculi in the cold room; (**c**) building section illustrating the proportions of the three bathing spaces; (**d**) roof plan with oculi location; (**e**) internal view of the blocked oculi in the warm room.

This organisation is found in the vast majority of the hammams (11 out of 13) as illustrated in the case of hammam Sidi Ayyub (Figure 5) and is reminiscent of the early Moroccan Islamic hammams excavated near the Roman settlement of Volubilis and the Alomravid settlement of Aghmat. In addition to the bathing spaces, all hammams comprise a changing room with an intermediate space with toilets connecting the changing room to the cold room. The configuration of three successive vaulted rectangular bathing rooms is reflected in the roof architecture of 11 out of the 13 heritage hammams surveyed as illustrated in Figure 6. The architecture of the cold and hot rooms, consisting of long rectangular rooms of varying dimensions, remains constant in all the surveyed hammams. The three rooms are covered by barrel and/or crossed vaults pierced by a number of roof openings.

**Figure 6.** *Cont*.

**Figure 6.** *Cont*.

**Figure 6.** *Cont*.

**Figure 6.** *Cont*.

**Figure 6.** Roof plans of the 13 surveyed hammams. (**a**) Hammam al suq (**b**) Hammam Bab Ailene (**c**) Hammam Ziani (**d**) Hammam Laksour (**e**) Hammam Sidi Ayub (**f**) Hammam Qannaria (**g**) Hammam Sidi Abdelaziz (**h**) Hammam Bab Doukala (**i**) Hammam Derb Arjane **(j**) Hammam Mouassine (**k**) Hammam Dahab (**l)** Hammam Azbezt (**m**) Hammam al Bacha.

The examination of the number and location of roof oculi eveals some recurrent patterns as follows:


larger hammams, as illustrated in the roof plans of hammams Mouassine and Doukala (Figures 7 and 8). The dome openings are based on the octagonal formation of the dome resulting from a 45-degree rotation of one square plan over another. This explains the number of dome openings as multiples of four.


**Figure 7.** *Cont*.

**Figure 7.** Hammam Bab Doukala from top left clockwise: (**a**) Roof plan; (**b**) interior view of hot room pierced vault; (**c**) warm room central pierced dome; (**d**) building section; (**e**) warm room pierced cross vault.

**Figure 8.** *Cont*.

**Figure 8.** Hammam Mouassine: From top left clockwise: (**a**) Roof plan; (**b**) warm room cross vault; (**c**) section across all the hammam spaces; (**d**) external view of warm room's dome; (**e**) internal view of the warm room's dome.

The comparative analysis of the plans of the 13 hammams reveals that the warm room and the changing area are the main spaces where more architectural variations and elaborate oculi configurations are to be found. This is clearly illustrated in the hammams of Bab Doukala (Figure 7) and Mouassine (Figure 8). Both hammams display the same principle of space and roof configuration in their warm room where the space is based on a central square configuration with a central pierced dome and a series of pierced cross vaults surrounding the central domed square. More elaborate architecture in their changing room is also found as illustrated in Figures 7 and 8. A systematic examination of the location of these two hammams in their wider urban context reveals their proximity to a large complex of a Friday mosque, madrassa and public fountain. Being part of a cluster of important public facilities at the wider urban scale, these two hammams enjoy a larger catchment area of bathers. Since the changing room and the warm room are the spaces where bathers spend most of their time, this explains their larger scale and more elaborate architecture. Furthermore, the changing room acts as a hub for bathers' relaxation and social interaction and as a venue for pre-wedding hammam celebrations for future grooms and brides. The majority of the hammams are, however, smaller structures with modest size and architecture and are located in the proximity of small residential neighbourhood mosques. The roof configuration as well as the number and location of oculi is therefore dependent on the size of the hammam, which depends on its location in the city within the medieval urban fabric's hierarchical system of clusters of facilities. Hammam al Bacha, a very large hammam built in the 1950s, is an exception due to its relatively recent construction.


**Table 2.** Floor area, number of roof openings and ratios of total roof opening area in the three bathing spaces of the 13 surveyed hammams.

*3.2. Analyses of Total Area of Roof Openings Expressed as a Percentage of the Internal Floor Areas of Each of the Three Bathing Spaces in the 13 Surveyed Hammams*

The internal floor area, the number of roof oculi, and the ratio (in percentage) of total area of roof openings in relation to internal space floor area have been calculated in each of the three bathing spaces of the 13 hammams as presented in Table 2.

The number of oculi in the roof of each of the bathing spaces is dependent on its floor area as illustrated in Figure 9a,b, where the graphs show (as expected) a clear correlation between the floor area of the space and the percentage of roof openings' total area in both the hot and warm rooms as illustrated in Figure 9a,b. The largest number of oculi is found in the warm room, when this has a central dome as is the case of hammam Mousassine (Table 2). However, the key finding here is that the cold room's total area of roof openings, consisting on one row of eight oculi, is not related in any way to its internal space floor area (Figure 9c).

As the cold room is the first space bathers cross on their way to the warm and hot rooms (where most of the bathing takes place), its low levels of daylight help bathers' sight to adjust to the low light levels in the steamy bathing spaces of the warm and hot rooms. This indicates an underlying principle of allowing bathers' sight to adjust to the lowest level of light in the cold room so that they ge<sup>t</sup> accustomed to the slightly higher light levels in the warm and hot rooms, where most of the bathing activities take place. The bathers' journey through different spaces of increasing or decreasing temperature and steam is matched with an increasing or decreasing amount of daylight. The darkest room is the cold room where the number of oculi is not dependent on its internal floor area. There seems to be some symmetry in the transitions between extremes conditions of hot and lighter, cold and darker. The warm room presenting conditions of transitions between the two extremes of cold and hot room.

**Figure 9.** Relationship between floor area and number of roof oculi per type of bathing space: (**a**) The hot room; (**b**) the warm room; (**c**) the cold room. The horizontal axis is the hammam as they appear in Table 1.

It is clear from Table 2 and Figure 10 that the cold room has the smallest proportions of roof openings as compared to the hot and warm rooms in the vast majority of the surveyed hammams. It is also interesting to note that the maximum total area of oculi roof openings does not exceed 2% of the internal floor area of the bathing space above which they are located as illustrated in 12 of the 13 surveyed hammams (Figure 10).

**Figure 10.** Comparison of ratios of roof opening total area in relation to floor area in the three bathing spaces of the 13 surveyed hammams.

### *3.3. Restoring the Vernacular Daylighting System and Quantifying Horizontal Illuminance Afforded by It in the Three Bathing Spaces of a Working Hammam*

All surveyed heritage hammams have completely lost the original daylight qualities in their bathing spaces, as the vernacular blown glass caps (Figure 11) have all been replaced by poor alternatives, such as glass bricks or flat sheets of reinforced glass (Figure 12a), or have been completely blocked by layers of cement. Poor incandescent or fluorescent electric lighting that is neither appropriate nor safe for the steam conditions of these spaces was found to be used all day long in the surveyed hammams, increasing consumption of electricity. Furthermore, face-to-face interviews conducted with three female hammam workers in Marrakech indicated the occurrence of headaches and fatigue which they attributed to the lack of daylight and natural ventilation in the bathing spaces. A follow-on discussion with the head of the hammam managers association in Marrakech revealed that there was a lack of knowledge of the original blown glass caps which used to cover the roof oculi amongs<sup>t</sup> most hammam managers. This situation is further exacerbated by the loss of glass blowing workshops in Moroccan cities, with the exception of the one and only glass blowing workshop opened by a French entrepreneur in Marrakech and employing the last two glass blowers in Morocco. Based on the form of vernacular glass bells, the transparent blown glass bells of 18 cm in diameter was used by the author as a prototype to produce locally 48 glass bells, using recycled glass (Figures 11 and 12).

**Figure 11.** Samples of blown glass bells produced in Syria in 2007 based on vernacular prototypes examined to restore hammam daylighting. The blue version is an openable oculus glass cap for natural ventilation.

**Figure 12.** The roof of Hammam Rjafalla: (**a**) Removing poor alternatives of oculi covers; (**b**) re-opening the oculi and placing the blown recycled glass caps; (**c**) solar panels over the warm room affecting indoor daylight measurements; (**d**) blocked oculi in the hot room; (**e**) vernacular daylight restored; (**f**) oculi natural light effect.

Access to a working heritage hammam in Marrakech to restore the blown glass caps over the oculi system was impossible. However, the opportunity to restore the vernacular daylighting system in a working heritage hammam presented itself in Rabat (the administrative capital of Morocco) in July 2016. Built in the 1950s, during the French protectorate, the Hammam Rjafalla is located in an informal, low-income housing neighbourhood, known as Hay Rjafalla, in the area of Yacoub al Mansour. The building presents a typical Moroccan vernacular hammam architecture, consisting of three successive rectangular parallel rooms with two rows of eight oculi in each of the three north/south facing vaulted roofs (Figure 12c).

The rehabilitation of the vernacular daylighting system took place in July 2016, under the initiative and supervision of the author, at a time when the hammam was temporarily closed for its annual maintenance work. The hammam manager has been in charge of the building for the last 36 years and is the only active female member of the National Federation of Moroccan hammam managers. She is also one of the key participants to the sustainable hammam initiative in Morocco which was presented at the COP22 in Marrakech in November 2016 [16]. The re-opening of the closed oculi and the installation of the blown glass bells was carefully conducted in order to avoid damaging the pottery tubes located in the masonry of the vaults and which form the base for the glass cap (Figure 12c). Once the vernacular daylighting system was restored, HOBO data loggers were installed in the same position within each of the three parallel bathing rooms at a height of two metres from the floor and at the base of the vault. This allowed for the recording of horizontal illuminance at a similar point on the internal south facing wall under the steam conditions of a working hammam and without interfering with bathers' activities (Figure 13).

**Figure 13.** Installation of HOBO data loggers for measuring temperature and light: Temperature range: −20 ◦C to 70 ◦C, light measurement range: 0 to 320,000 lx (0 to 30,000 lumens/ft2).

The data loggers were set to record light levels simultaneously in the three bathing spaces, under the working hammam heat and steam conditions and with the oculi openings being the only source of daylight. A fourth HOBO data logger was placed on the hammam roof (on top of the warm room) to measure the horizontal illuminance on the roof, simultaneously with the data loggers installed inside the building. Measurements were taken from 19 July to 30 September 2016 at 20 min intervals.

The plotting of weekly averages of horizontal illuminance (recorded every 20 min between 11:00 a.m. and 17:00 p.m.) over a period of almost two months and a half, clearly indicates that the averages of daylight levels do not exceed 25 lx and that the cold room tends to have slightly higher averages than the hot room despite their same configuration and orientation (Figure 14). However, when examining daily plotting of recorded illuminance in the three spaces, the results show that horizontal illuminance (lx) recorded in the hot and cold rooms varies between 10 and 60 lx with the hot room recoding slightly higher levels than the cold room (Figure 15 and Table 3).

**Figure 14.** Weekly averages of horizontal illuminance (lx) recorded in the hot and cold rooms (19 July to 30 September 2016).

**Figure 15.** Horizontal illuminance values recorded in the three bathing spaces during the summer of 2016 on a daily basis between 11:00 a.m. and 05:00 p.m.


**Table 3.** Weekly averages of horizontal illuminance recorded in the cold room, the hot room and the room.

The plotting of horizontal illuminance levels over a period of two months and a half clearly indicate that it is in the hot room where the highest levels of horizontal illuminance were recorded (Figure 15). However, these levels remain well below 100 lx and do not exceed 60 lx (Figure 14). The levels recorded for the warm room remain the lowest as they have been jeopardised by the installation of two large solar panels, casting shadows over the oculi (Figure 12c) and resulting in levels being at 10 lx or below despite the sunny and clear sky conditions during the summer months. However, this indicates that the vernacular daylighting provision works best in conditions where the roof vaults have a good sky exposure and are not shaded by roof installations or adjacent buildings.

There is a rapid decline in weekly illuminance averages in September leading to much lower levels inside the bathing spaces (Figure 14).

The plotting of the 21 July 2016 measurements in the cold, warm and hot room, as well as on the roof, reveals a more detailed pattern of illuminance variation in the three bathing spaces and the roof sky conditions (Figure 16). These plots, together with Table 3, illustrate a clear correlation between the sky conditions and the daylight levels afforded by the vernacular system.

**Figure 16.** *Cont*.

**Figure 16.** Horizontal illuminance in the three bathing rooms and on the roof and the calculated daylight factor on the 21 July 2016 between 11:00 and 17:00.

The drop in light levels is much more sudden in the hot room than in the cold room and this could be explained by the high level of steam (98% humidity) reached in the space by early afternoon, when the use of the hot room by bathers is at its highest. The daylight factor calculation for the same day, based on real data, reveals that there is a sudden reduction in the DF in the hot room early afternoon (Figure 16b) below that of the cold room as the humidity and steam in the hot room reach saturation levels.

The comparison of recorded illuminance on the roof and the three bathing spaces on the 21 July 2016 (between 11:00 and 17:00) clearly illustrates that the sky conditions are directly reflected into indoor daylight levels in the hammam bathing spaces (Figure 16). This indicates that the vernacular daylighting system in hammam buildings allows for users' strong connection to outdoor sky conditions and to the movement of the sun in the sky which is known to be important for users' well-being.
