1. Context and concept

Within the framework of architectural and historical studies, the relevance of dating techniques has been illustrated through performance prediction of existing structures, as well as by the selection of appropriate replacement masonry units (Laefer et al., 2004). In addition, as far as the conservation of built heritage is concerned, building dating helps specialists in the field of cultural heritage to justify conservation or reconstruction measures with respect to the historical and aesthetic values of a monument.

Built heritage is dated, whether absolutely or relatively, by using historical sources, such as ancient maps and plans, drawings, paintings and written records that testify to the year of construction or transformation. The archaeology of a building is also useful to confront results obtained from historical studies and helps to better understand the morphological transformations of a construction through time. Laboratory tests may also provide data regarding material properties and compositions. Such characteristics, regarding the type of material used and their specific characteristics, may be related by specialists to a particular period of time or region of production. Where ceramic materials are concerned, these laboratory tests require samples that can be analysed with expensive and sophisticated dating tools, such as the thermoluminescence technique (TL), the optically stimulated luminescence technique (OSL), or radiocarbon and archaeomagnetic dating (Stella et al., 2014). Such studies are always time-consuming and never operated on-site, which also draws a barrier between the archaeologists and scientists.

In this context, the primary motivation of this study was to evaluate the validity of an innovative method for dating monuments with low-tech, non-destructive instruments, which are easy to handle, even for non-scientists. The second objective was to estimate the correlation between the decay of a brick wall exposed to weathering and the date of construction of the monument. Among the non-destructive tools used to assess the hardness of building materials and the effect of climate on their resistance, the Schmidt rebound hammer is one of a kind.

Initially used for concrete hardness characterisation, the Schmidt hammer was, at its origin, employed for its ability to quickly assess the strength of concrete through the value of its mean rebound values. Since then, this method has also been applied to rock and ceramic materials. Based on such a method, its use was soon diverted to rock studies as a dating method known as Schmidt hammer exposure dating (SHED). This method is based on the principle that a rock surface which is uncovered for longer periods of time is relatively softer compared to a covered one. Since this hardness is measurable with the hammer, it becomes possible to build a calibration curve based on different measurements from stones of known ages. Rock-surface hardness as an indication of exposure has also been used in archaeological contexts, as it is non-destructive, designed for in situ testing and has proven to be an extremely efficient technique for field assessments (Betts and Latta, 2000). This field of research is still being explored in geological studies in which the surface weathering of a rock outcrop is widely used as an indicator of rock age (McCarroll, 1994; Shakesby et al., 2006; Sanjurjo et al., 2009; Stahl et al., 2013; Matthews et al., 2016; Tomkins et al., 2016; Tomkins et al., 2018). Research carried out by Tom-kins et al, (Tomkins et al., 2016) indicates that granite weathers linearly over significant spatial scales for regions with a similar climate. According to Shakesby et al, (2006), the weathering rate declines slowly over very long timescales, often exceeding 100,000 years. Although these studies provide SHED calibration curves that correlate with the degree of decay of a rock type with its age, the time laps under the scope are measured in kilo annum (ka). Therefore, the validity of SHED for dating monuments is a central question to this research, as very few studies have explored this domain. The use of the Schmidt hammer for assessing the state of decay of a material was discussed by Iqbal (Iqbal, 2016). She proposed a new approach for quantifying the soiling and dating exposure of limestone façades based on rebound values, for time laps that do not exceed 30 years of exposure to the environment. Her research showed that the higher the level of contamination or soiling accumulation, the lower the surface hardness. Similarly, the results presented in this paper are based on the assumption that historical brick masonry which is exposed to climatic conditions for a longer period of time is likely to be more weathered in comparison with the earlier brick masonry.

2. Historic Bricks

Bricks produced prior to the mid-20^th century show high variability in colour, geometry and mechanical properties (Laefer et al., 2004) compared to the industrially made ones. This results from the quality and variability of the raw material, as well as the traditional techniques and methods used for shaping and firing. In this paper, the traditional bricks which are studied are differentiated from the industrial ones, as they are made by hand and differ by the methods used for shaping and firing the raw material. Hand moulding and pressing were in use for manufacturing fired-clay bricks before mechanical extrusion appeared. The firing process also involved the use of a kiln type that saw some technical improvements over time (Perez-Monserrat et al., 2017). The firing technique influences the physical and chemical properties of the product, as well as its weathering resistance to climatic conditions, as the firing temperature is a major factor that influences the degree of vitrification and the brick’s resistance to decay (Elert et al., 2003; Cultrone et al., 2004). On this particular aspect, it is well known that bricks fired at higher temperatures have higher durability. A longer heating period will also ensure a better-quality product.

Among the different types of kilns used to fire bricks, the field kiln was the most primitive and popular, and was used from the early ages until the 19^th century. It was built near the construction site for temporary use only. This type of kiln is erected with raw bricks, stacked together and covered with a layer of clay and grass. During firing, the whole production is exposed to varying temperatures, depending on the relative position of the bricks within the kiln, which in turn affects the mechanical properties, porosity and water absorption of the material. This was proved by Laefer (Laefer, 2001 and Laefer, 2004) who showed that the variability from a single firing exhibited over a 100% strength difference between the strongest and the weakest brick.

The downdraft kiln is a refined version of the updraft model (see below) for larger production (50,000−60,000 bricks) and with better heat distribution. The flames surround the bricks, allowing the heat directed upwards to go downwards by means of flues placed under the kiln floor and connected to a chimney. Although this technique was an evolution in the brick-making process, it still gave a variable quality product which could influence the durability of the material. The continuous draft model is made of chambers with multiple firing places to allow the distribution of heat from one chamber to another by flues. This model was widely used until the mid-19^th century, as it reduced heat wastage. In 1858, Hoffman and Hieut improved the chamber kiln into a series of continuous chambers adjoined to each other in a circular form. Firing and cooling occur sequentially in the chambers, allowing the heat to be transferred from one chamber to the next for preheating. From 1910, continuous chamber kilns quickly became established in Europe, with 90 to 95% of all bricks burned with this technique. Similarly, the modern tunnel kiln invented in 1751 is based on the chamber kiln process. Here, brick production is moved on tracks over three days through a stationary heat source before cooling. The benefit of this evolution is mass production and reduction in maintenance costs. More details regarding kiln types, their history and performances can be found in Hoehne (1910), Rhodes (1968), Laefer (2001) and Laefer et al. (2004).

In the framework of this study, the author assumes that the bricks used for the tests were fired by updraft and downdraft kilns, as the churches used as case studies were erected between 1600 and 1895. Therefore, it was expected that the quality and properties of the bricks fired since the mid-18^th century until 1858 vary because of the joint presence of field kilns and tunnel kilns in the same area.

3. Methods

Sampling procedure

Ten churches, built with traditional bricks between 1600 and 1895, listed as built monuments and precisely dated by historical sources were chosen in a particular area of Belgium (Figs. 1 and 2). These monuments have the advantage of being mostly oriented, meaning that the altar usually faces east, whereas the main entrance faces west. This repetitive orientation of the buildings facilitates comparisons between measurements done on façades. Fig. 3 illustrates three churches built in 1717, 1725 and 1856 that were selected for the tests.

Fig 1

Map of Belgium and localisation of the region of Hainaut.

Fig 2

Localisation of samples within the region of Hainaut.

Fig 3

Church 2 (1717), Church 3 (1725) and Church 9 (1856).

Schmidt hammer tests were done on each façade of the buildings according to the points of the compass. For each façade, depending on the building’s scale, at least 20 bricks were selected, all located in two different zones of each façade. All tests were performed by the author with a Schmidt hammer type LB held horizontally, perpendicular to the brick surface during summer time (T_Mean = 15.4°C; RH = 69%). Each brick was successively impacted five times at its centre, and for each impact done on each brick, a rebound value R_n was calculated for each façade.

4. Results and discussion

Rebound values

Table 1 provides the mean rebound values measured for the first five impacts R_n (with n=1−5) for each building and each façade. No measurements were acquired for the west façades of Churches 1 and 2, nor were they for the east façade of Church 10, due to the occurrence of stone masonry at these locations.

Table 1

Rebound values measured for the first five impacts, R_n (with n=1−5).

Samples	Date	Orientation		Mean R values				Mean R1(SD)	Mean R5 (SD)
			1	2	3	4	5
Church 1	1600	North	29	32	35	36	35
		South	23	30	35	34	34	25.0	34.4
		West	/	/	/	/	/	(3.1)	(0.4)
		East	24	29	32	33	32
Church 2	1717	North	26	31	37	35	37
		South	25	29	32	33	32	24.8	34.1
		West	/	/	/	/	/	(0.8)	(2.5)
		East	24	28	29	33	33
Church 3	1725	North	29	34	37	37	39
		South	28	32	35	36	39	26.3	37.1
		West	24	30	33	34	34	(2.5)	(2.3)
		East	24	31	33	34	36
Church 4	1769	North	27	34	38	38	39
		South	31	38	40	39	42
		West	30	34	35	37	38	29.5	40.0
		East	29	39	38	41	42	(1.7)	(1.8)
Church 5	1774	North	32	38	40	42	43
		South	33	38	40	41	41	31.2	41.1
		West	31	37	40	41	41	(1.3)	(1.7)
		East	30	36	38	38	39
Church 6	1782	North	28	34	37	40	40
		South	28	35	39	40	41	27.3	39.6
		West	25	31	35	35	36	(1.3)	(2.7)
		East	28	36	37	39	41
Church 7	1790	North	28	32	34	37	37
		South	21	27	33	35	37	24.2	36.1
		West	25	33	36	36	36	(3.1)	(1.6)
		East	23	29	33	36	34
Church 8	1840	North	24	31	33	34	36
		South	22	29	32	35	36	25.5	36.9
		West	30	37	38	39	42	(3.2)	(3.2)
		East	26	32	35	34	34
Church 9	1856	North	24	31	34	36	34
		South	27	31	33	36	35	26.3	35.1
		West	25	29	31	31	34	(2.3)	(1.6)
		East	29	35	35	36	37
Church 10	1895	North	31	34	36	38	38
		South	29	32	35	35	37	29.6	37.7
		West	28	34	34	35	38	(1.6)	(0.9)
		East	/	/	/	/	/

The plot of the mean Schmidt hammer rebound values R₁ and R₅, built with all the measurements from the façades and compared as a function of the construction dates is shown in Fig. 4. The results indicate that the mean R₁ rebound values are within the range of 24.2 (Church 7, 1790) and 31.2 (Church 5, 1774), while the mean R₅ values are within the range of 34.1 (Church 2, 1717) and 41.1 (Church 5, 1774). Although the oldest monuments show the lowest R₁ values, the results also show that the different R₁ values do not strictly increase from the oldest building (Church 1) to more recent ones, at least until 1790. From 1717 (Church 2) to 1774 (Church 5), the mean first rebound value increases before dropping to lower values. Considering the results of the whole dataset, the hardness of the bricks is not corelated with the age of all the monuments. This suggests that other parameters, such as raw material quality and the manufacturing process, influence the conservation of the building material.

Fig 4

Mean Schmidt hammer rebound values R₁ and R₅ as a function of the construction dates. Uncertainty bars are shown.

A clear correlation exists between R₁ and R₅ values as both the mean values fluctuate in the same proportion for the time laps with a ratio of 127% (minimum, Church 10) and 149% (maximum, Church 7). The mean R₅ values for all the buildings indicate that the material strength is not higher for more recent products. On the contrary, bricks used between 1769 (Church 4) and 1782 (Church 6) present higher R₅ values, which could indicate that these products are more resistant compared to the other samples. Fig. 4 shows that hard bricks with high R₅ values, such as those tested on Church 5, also present higher R₁ values. Therefore, there is a clear dependency between the material strength and its resistance to weathering. In other words, the hardness of the brick influences the degree of decay of the material when it is exposed outdoors. The most ancient bricks dated back to 1717 present lower R₁ rebound values, which confirms that the oldest brick samples were more weathered than the more recent ones.

Contrary to what could be expected, the R₅ mean rebound values did not linearly increase through time. Although the R₅ mean hardness values were relatively stable from 1790 to 1895, extreme R₅ values were measured for churches between 1769 and 1790 (Fig. 4) which may indicate that both the field and tunnel kilns were in use since the mid-18_th century. All in all, the expected age-hardness relationship was not observed for the whole dataset. Churches built between 1790 and 1895 showed increasing R₁ values, which indicate the effect of weathering on these samples, while R₅ values, representative of brick strength, tend to stabilize, showing products of comparable strength.

Weathering indices

A plot of the weathering indices (I₅) as a function of the building dates is shown in Fig. 5. For better understanding, the results have been grouped into four orientations, corresponding to four points of the compass (Table 2).

Fig 5

Weathering indices (±95% confidence intervals; n=30) as a function of the building dates and grouped into the orientations of the compass.

The results indicate that I₅ (%) values do not strictly decrease from 1600 to 1895, which is coherent for the first and fifth rebound values discussed in the previous section. However, things are different for churches built after 1790. It is interesting to call to mind that these samples also present increasing R₁ values. Variations of I₅ (%) values (Table 2) are contained within a range of 0.31% (Church 6, South/East) and 19.33% (Church 10, North/East).

Table 2

Weathering indices (I₅) as a function of the building dates and specific orientations of the façades (mean ± 95% confidence interval; n=30).

		Church 1	Church 2	Church 3	Church 4	Church 5	Church 6	Church 7	Church 8	Church 9	Church 10
Date		1600	1717	1725	1769	1774	1782	1790	1840	1856	1895
	South/West	32.9 ± 2.5	23.1 ± 3.6	28.8 ± 0.1	22.9 ± 1.3	22.5 ± 0.9	30.9 ± 0.9	36.0 ± 3.0	28.3 ± 2.5	24.8 ± 0.9	22.4 ± 1.4
15(%)	North/East North/West	24.6 ± 3.6 17.8 ± 0.5	29.7 ± 0.9 30.8 ± 4.2	29.4 ± 2.2 26.8 ± 0.9	30.0 ± 0.2 25.9 ± 2.7	25.7 ± 0.6 25.9 ± 0.6	31.0 ± 0.5 29.6 ± 0.3	27.9 ± 1.7 28.5 ± 1.9	28.9 ± 2.3 31.0 ± 1.3	25.7 ± 2.3 28.2 ± 0.9	18.1 ± 3.5 20.7 ± 2.0
	South/East	32.2 ± 0.4	24.5 ± 1.1	31.4 ± 1.3	27.3 ± 1.2	22.3 ± 0.8	32.2 ± 0.1	35.6 ± 3.2	32.5 ± 3.5	22.3 ± 0.5	22.4 ± 4.2
Mean 15 (%)		27.4 ± 3.1	27.1 ± 1.4	29.2 ± 1.3	26.3 ± 1.8	24.1 ± 0.9	30.9 ± 0.6	33.0 ± 2.9	30.9 ± 2.2	25.2 ± 1.4	21,4 ± 1.5

Fig. 6 shows the relationship between I₅ (%) values with the corresponding age of the construction. The dataset can be divided into three groups (G1, G2 and G3) with specific observations. Group G1, constituted of Churches 1 to 3 (constructed between 1600 and 1725) shows rather constant weathering indices for the whole period of time, although

Fig 6

I₅ (%) values with the corresponding age of the construction (G1, G2 and G3, see text).

only few buildings of this period were found. During a very short period of time (1769−1790), a variability of I₅ (%) values was observed for group G2 (Churches 4 to 6). Within this period, the lower I₅(%) values may correspond to better products and bricks fired at higher temperatures. Logically, bricks presenting higher weathering indices could be viewed as potentially more altered because of poorer quality. Group G3 concerns churches built since 1790, with I₅(%) values with decreased weathering indices over time.

For this group (G3), the equation of the best fit line and the coefficient of determination (R²) are listed in Table 3. A high coefficient of determination, up to 0.8, was found for each orientation of façade, except for north−west façades (R²= 0.6). The best result, with the maximum coefficient of determination, is obtained when a mean value of I₅ (%) is calculated, based on all the façades‘ measurements.

Table 3

Coefficients of determination for each orientation of façade.

Orientation	Formula	R2
South/West	Y = 0.1872x+372.84	0.903
North/East	Y = 0.1903x+377.64	0.845
North/West	Y = 0.0867x+185.35	0.637
South/East	Y = 0.0911x+192.27	0.863
Mean	Y = 0.1484x+299.76	0.948

5. Conclusions

Fired clay bricks are well known to present non-heterogeneous structures, mainly influenced by the raw material, the firing process and the long-time exposure to climatic conditions.

The presented study was the first step to explore the occurrence of a relationship between the degree of decay of brick masonry and its age. A non-destructive tool able to evaluate the harness of ancient bricks was used to compare the results obtained from a sample of 10 churches, built between 1600 and 1895.

Schmidt hammer tests were carried out on each façade in order to measure the mean rebound values (R_n with n=1−5) for bricks. These tests refer to the resistance of the surface to successive n impacts of the hammer on the brick surface. Based on these results, for each façade, the weathering index I₅ (%) was used to quantitatively evaluate the degree of decay of the brick surfaces.

The tests indicate that the Schmidt hammer is efficient for evaluating historical fired clay brick weathering, as the rebound values measured for each successive impact were significantly affected by the degree of decay of the material exposed to climatic conditions for centuries.

However, the results illustrate that the first and fifth mean rebound values do not decrease with the respective time of construction for the whole dataset. This means that R₁ and R₅ values are not strictly correlated with the ages of construction. Therefore, this research confirms that other factors, such as the quality of the raw material (composition and homogeneity), the type of kiln used for firing, as well as the parameters that define the process (brick locations within the kiln, temperature, duration) play an important role in the capacity of the material to resist decay when it is exposed to identical climatic conditions.

The mean R₅ values and mean R₁ values measured for the whole data set vary in the same proportions, which indicates that the first rebound measurement is a reliable indicator for estimating brick quality. However, for the samples under scope, the research also highlights that there is no relationship between the quality of brick and its age. This also means that older bricks can show higher hardness and higher compressive strength, compared to recent samples.

The fact that the mean R₁ values were strongly correlated with R₅ ones, and that superficial hardness is not strictly related to the time of exposure, also explains why the weathering indices I₅ (%) values did not increase with the age of the building. Indeed, this rule was only observable for buildings constructed between 1790 and 1895. This trend could be explained by better-quality products, as confirmed by the low fluctuation of mean R₅ values, during this period. This hypothesis is also confirmed by the invention of modern kilns which appeared in 1750, and the progressive abandonment of the traditional kilns in the area under scope. The difficulty to establish a strong relationship between the date of construction and decay of the bricks was also illustrated by the equable I₅ (%) values measured for the oldest samples dated between 1600 and 1725. Although only three samples are concerned, for such a period of time of exposure, it is unconceivable that decay remains constant, contrary to the indications of the samples spread out between 1769 and 1790.

Based on these observations, the preliminary results tend to indicate that SHED is a hazardous technique to date masonry dating back to 1790 or that made of bricks fired with uncontrolled methods and traditional kilns. This is because the Schmidt hammer measures the hardness of the material and this hardness depends not only on the weathering factors but also on the initial properties of the product. As such, it is difficult to interpret the respective influence of both parameters.

However, the occurrence of a high correlation between the parameters, with a coefficient of determination R²ranging between 0.64 (north/west) and 0.90 (south/west), with a mean value of 0.95, indicates that SHED is a valid approach for masonry buildings that have been standing since 1790. The influence of the orientation of the façade on the R²values was also highlighted and shows the influence of sampling when conducting SHED tests on masonry. Such findings reduce expectations for older constructions. Future works should firstly confirm these preliminary results by enlarging the dataset with more samples. Hypotheses related to the influence of the brick-making process will be examined through the scope of complementary laboratory tests carried out on samples (petrographic analysis, mineralogical composition and micro-structure), as well as through deeper historical studies related to kiln technology and evolution.