Journal of Archaeology in the Low Countries 4-2 (April 2013)Martijn van Haasteren; Maaike Groot: The biography of wells: a functional and ritual life history

4. Ritual deposits in wells

It is generally accepted by archaeologists that ritual deposits occurred in wells. Several rich deposits in wells have been interpreted as ritual, such as half a Roman helmet with an iron ploughshare in Breda, and two wagon wheels with 27 amber beads at Forum Hadriani/Voorburg (fig. 6; Hoegen et al. 2004, 366; Bink & Franzen 2009, 93-95). These deposits were interpreted as ritual because no functional explanation could be given. A complete ceramic pot or wooden tool are also special finds, but are not automatically interpreted as ritual, since these are considered less valuable. By collecting data on ritual deposits and other finds from wells, it should be possible to recognise less obvious ritual deposits, for instance consisting of finds that are understood to be of little value. In this paragraph we hope to recognise any recurring patterns within deposits, by grouping them in different ways. The study that this paper is based on described around 60 deposits in detail (Table 1; Van Haasteren 2011, 25-40). Before going into the actual grouping of deposits we will explain how this data set was collected.


Figure 6 Section through a well in Voorburg, showing two wagon wheels. The amber beads were found between the wagon wheels ( Bink & Franzen 2009 , fig. 9.31 & fig. 5.9; adapted by B. Brouwenstijn).

4.1 Inventory

next section

To make our inventory reports about excavations of Late Iron Age and Roman period sites in The Netherlands have been searched for information on finds from wells. Besides libraries and lists of publications on websites of archaeological companies, the e-depot for Dutch archaeology ( has been an important source for these reports.

Naturally, the deposits were not picked randomly from these reports. If the authors of a report marked a find as ‘special’ or ‘ritual’, it was automatically included in our selection. The selection was expanded by including what we understood to be ‘special’ finds. The criteria for selection are mainly comparable to the ways for identifying ritual deposits discussed above. In the first place deposits with remarkable or valuable objects have been selected. Secondly, deposits with wooden objects and complete ceramics have been added, as well as remarkable deposits of bone, and stone objects. Besides special objects, attention is paid to deposits in which various materials are found together. Finally, the location of the deposit has been an important focal point. As has been said before, to find objects in one part of a well can be more exceptional than in other parts. Furthermore, it is important to note that because of these criteria deposits of fragmented objects have rarely been selected. That does not imply, however, that they cannot be (part of) ritual deposits.

Something that also played a role in the selection of deposits is the way in which wells and deposits have been described in reports. Some archaeological services produce more elaborate reports than others. When features or finds are only briefly mentioned or listed in tables, then the possibility that they were included in our selection decreases. On the other hand, if finds were described in detail, and information on other finds and the find location within the well was included, they were far more likely to be included in our selection.

This inventory focused on identifying potentially ritual deposits from a range of sites in The Netherlands. In the future, a more extensive inventory of all published wells from a selection of sites may provide a useful indication of the extent of special finds in wells.

4.2 Grouping of deposits

The deposits have first been grouped according to their material category, since this is usually the way finds are presented in archaeological reports. Moreover, synthetic studies of ritual deposits often focus on one category, such as animal bone or bronze (Groot 2008; Fontijn 2003). As we shall see, the danger involved in splitting up finds according to material properties lies in losing sight of similarities between finds. Depositions of the following materials have been identified: pottery, stone, animal bone, metal, wood, leather and botanical remains. It is mainly rich deposits and deposits of combinations of different types of material that are interpreted as ritual deposits in archaeological reports. Apart from the examples already described above, other ‘rich’ deposits are a ceramic jug found inside a copper cauldron in Geldermalsen-Hondsgemet, two bronze vessels with a quernstone and a wooden animal head from Voorburg, and a concentration of animal bones with a brooch from Heeten (fig. 7; Van Renswoude 2009, 271; Bink & Franzen 2009, 107-108; Kenemans & Van der Velde 2007, 176).


Figure 7 Section through a well in Voorburg showing a quernstone and two bronze vessels in the lower fill, including the bronze bowl ( Bink & Franzen 2009 , fig. 8.43 & fig. 5.21; adapted by B. Brouwenstijn).

There are several arguments for a ritual interpretation of two find categories usually explained in functional terms: wooden ladders and spades. First, some of the ladders were placed in the well upside down. Second, one complete ladder is much too short to have been of any practical use with regard to cleaning the well. Third, a ladder in Midlaren was found together with another special find: a wooden bow. Finally, in an Early Medieval well in Raalte, three spades were found next to each other, indicating deliberate placement (Kooistra & Van Haaster 2001, 327-332; Verwers 1992, 174; Nicolay 2008, 168; Bloo et al. 2007, 190-191).

A second grouping is based on the function of the objects, and distinguishes four categories: containers, tools, clothes and personal items, and food. The aim of this classification was to look at the finds from a different perspective, independent of the used materials. One of the discoveries was that drawing water is unlikely to have been the primary use of the containers found in wells. Cooking pots, salt containers, pots containing cereals, wicker baskets and a wooden bowl have all been found in wells (Hoegen 2004, 255; Vos 2002, 59-60; Sier 1999, 67; Hiddink 2005, 288-291). Self-evidently, a wicker basket would be useless for holding water, and pots with contents were also clearly not used for drawing water. While these could represent waste (e.g. moldy grain), this would only make sense if the well had already gone out of use. Tools include the ladders and spades, which have already been mentioned, as well as ploughshares, which were found in several wells (fig. 8). Even when damaged, ploughshares could have been remade into other objects, so they are unlikely to represent waste. The category food includes deposits of animal bones and plant remains. An example is a deposit of three rare medicinal plants from Helden; the plants are said to have had a ritual significance among the Germans (Van der Meer 2008, 9).


Figure 8 Ploughshare and helmet from a well in Breda ( Koot & Berkvens 2004 , fig. 14.9 & 14.10).

A problem with a non-material classification is that there are many different ways to define categories. For instance, ‘containers’ could be seen as part of a category ‘kitchen and cooking utensils’, which would include deposits with quernstones. Food could also be seen as part of a wider category. Bradley discusses deposits of ‘animated materials’, which are objects that contained life (Bradley 2005, 52). It is indeed interesting to separate this category of ‘animated materials’. Apart from animal bones and seeds and fruits, human bones and (unworked) wood also belong to this category. No examples were found of deposits of human bones in wells during our inventory. Examples of deposits of wood are a piece of maple wood together with two ceramic pots, a piece of oak wood, and an oak plank with the skull and front legs of a horse (Hoegen 2004, 257-258; Van Putten & Ter Wal 2006, 37-38; Groot 2009b, 398). The ‘flow stakes’, which have been described earlier, may also be attributed to this category, if it is indeed true that they did not improve the flow of water.

A final way of classifying deposits is according to their location within the well, and as such, their relation to its lifecycle. Within a well, finds can be found in different locations, such as on the bottom, in the fill or top fill, but also in the construction pit or deconstruction pit. Some types of materials are more likely to be found in some locations rather than others. Finds from the construction pit of a well are less likely to represent rubbish, since the chance of rubbish entering the construction pit are small due to the short time period in which a well is constructed. It is more likely that refuse ends up in the top fill, because of its effect as an artefact trap. The locations which we have distinguished are – in order of the life cycle – the construction pit, the bottom of the well, the fill, the deconstruction pit and the top fill.

Examples of deposits in the construction pit are a Roman helmet found with an iron ploughshare (fig. 8), a leather shoe sole, a wicker basket and three deposits of a wooden spade. These spades show different degrees of wear. On the bottom of wells, deposits were found of several complete pots (fig. 9), pots with cereals, ladders, and several pieces of wood, such as a heavy pointed beam with wooden pins (Nicolay 2008, 154). The deposits of vessels seem to form a pattern. Although in some cases it can be ruled out that they were used for drawing water, we should still consider this explanation for other cases. However, if the ladders were indeed used for cleaning wells, and this task was frequently carried out, we should ask ourselves why these vessels were not recovered.


Figure 9 Section through a well in Breda with two complete ceramic pots in the lower fill ( Koot & Berkvens 2004 , fig. 11.44).

Examples of deposits within the fill include a heavy tuff stone pedestal, a concentration of pottery including a storage vessel containing the bottom of a jug, wagon wheels with barnstone beads, a concentration of bones from six cattle together with skulls of a horse and a ram, several deposits of complete eared pots with remains of rope in combination with manure, bone and pottery, and two large concentrations of wood (Van der Kamp & Polak 2001, 22; fig. 5; Bink & Franzen 2009, 87, 93-95; Groot 2009b, 401-4; Kenemans & Van der Velde 2007, 176; Waldus 2000, 37-38, 42-44; Niekus 2002, 20-21). Only one deposit is known to have come from a deconstruction pit. This is the find of two bronze brooches from Geldermalsen-Hondsgemet (Van Renswoude & Roessingh 2009, 592-595). The fact that the deposit consists of two brooches is an argument for a deliberate deposit and against accidental loss. Losing two brooches at once is unlikely to have gone unnoticed by the owner.

The final category consists of deposits in the top fill. These fills often contain many finds, mostly small fragments of rubbish, since such depressions function as artefact traps. Small, isolated items are therefore difficult to interpret as either rubbish or a ritual deposit. However, some finds are clearly deliberate or non-rubbish, such as a small complete amphora, a complete dog skeleton, the bottom part of a hand quern and a group of four complete dog skeletons (Berkvens 2004, 136-137; Groot 2009b, 401-403; De Wit 2000, 18-19; Kooistra 1996, 134-135, 180-181).

Possible patterns consist of complete pots on the bottom of wells and dog skeletons in the top fill. However, these patterns can also be explained from a functional point of view, as lost containers for drawing water, and the convenience of using an existing depression to bury a dead dog. It is details such as the unsuitability of some of the containers for drawing water and the presence of ritual deposits in the fill of wells with dog burials at the top that lead to an alternative conclusion of these finds as ritual deposits. Many of the finds from the fill also seem to be deliberate deposits of material.

The first method of classifying deposits -- by material -- has demonstrated that it is usually deposits of remarkable items (helmet, ploughshares) or of combinations of various items that can be interpreted as ritual. The classification by function has shown that less remarkable objects, such as containers, as well as items which are valueless, such as wood and twigs, were also used in ritual deposits. This is based on recurring patterns or the lack of a functional explanation for the presence of objects in the well. What did not become clear from these classifications was the reason behind the deposits. It is the location within the fill that tells us the moment of deposition with regard to the life of the well, and so indirectly the reason behind the deposit. For instance, deposits found in the construction pit can only have been placed there during the construction of the well. Finds from the bottom of a well relate to the period of use, while deposits higher in the fill have either been placed there when the well was no longer in use, or at the moment when the period of use was ended. The next paragraph will discuss deposits in relation to the lifecycle of wells.