Many show caves are routinely cleaned to remove dust, lint and other debris introduced or mobilised by visitors. A number of methods have been utilised with high pressure cleaning probably being the most common method. This paper investigates the impacts of high pressure cleaning on a flowstone surface in Imperial Cave at Jenolan. Surface lowering and textural changes were examined using a microerosion meter and scanning electron microscopy before and after washing. A surface lowering rate of about 0.018mm per wash was determined.

Although the experiment did not precisely mirror cave cleaning practice it does demonstrate that repeated passes of high pressure water do produce changes to the surface of flowstone with most of the effect being due to mechanical attack. Some recommendations as to washing practice are made.


Heavily used tourist caves require cleaning to remove dust, lint and other organic debris from walls and floors (Bonwick and Ellis 1985). The debris itself has a variety of deleterious affects on the cave environment. Cleaning methods used at Jenolan and elsewhere in the world have included application of water at low pressure and high volume, with and without brushing or scrubbing; at high pressure and low volume and occasionally steam cleaning and the use of surfactants. All of these methods can be expected to have adverse impacts.

Cave cleaning is destructive - as this study will show - but often necessary. Any objective study of the damage will incur further impacts. The Jenolan Caves management has had the longest history of active cave cleaning in Australia (Bonwick and Ellis 1985, Newbould 1976) and recently Jenolan Caves Reserve Trust has expressed concerns about the level of cave cleaning impacts. This concern is enshrined in the formally adopted plan of management for the Reserve. The Trust has accepted that assessment of the effects of cave cleaning will require further impacts. They allowed us to carry out an admittedly damaging experiment to gain some objective measurements so that recommendations for future washing practices could be made.

The microerosion meter (MEM) has been widely used in a variety of natural and cultural environments to assess current weathering rates (Spate et al 1985, Trudgill et al 1989). Applications have included estimation of erosion rates from limestone solution, coastal processes, wind and chemical erosion in arctic regions and in cultural situations such as on buildings and on Aboriginal rock art.

In this study the MEM was used to measure lowering of speleothem surfaces before and after high pressure washing. These measurements were backed by Scanning Electron Microscopy (SEM) of surfaces and washed-off debris.

The SEM technique has found an increasing number of applications within the field of geomorphology. In weathering studies it is used to study how processes operate on the scale of individual crystals (McGreevy 1985). Fracture patterns on individual quartz grains observes by SEM have been used for environmental discrimination (Krinsley and Doornkamp 1973). It is a technique which has also found wide applications in the conservation of buildings and monuments, particularly in Europe. Conservators have used SEM to study the effects of acid rain deposition on building stone (Lewin and Charola 1978). This is a particular problem in heavily polluted urban centres, causing disfigurement of buildings and monuments. Acid deposition also causes damage to the building stone itself such that buildings must undergo periodic cleaning. Much research has been devoted to methods of cleaning building stone and the damage consequent on cleaning. Here again the SEM has been a particularly useful method of analysis of damage at a microscopic scale (Amoroso and Fassina 1983, Viles 1990).


The Site

Site selection was difficult. We needed a site where SEM samples could be easily obtained without contaminating the surface by their very collection; where MEM sites could physically be installed - and without contaminating the dirty surface to be measured; where wash-water samples could be collected and where the necessary electric power and water could be supplied to carry out the washing process.

A large number of locations were examined within the Jenolan Caves tourist system to find a site which had not been washed by any methods in recent time, had been exposed to very many tourists, was not exposed to the public, from which it was possible to catch the runoff from washing and from where it would be possible to remove samples for examination by SEM. Meeting all these criteria proved very difficult but a site was finally selected in the Crystal Cities chamber of the Imperial Cave. The site had been washed previously (in 1981) and was not a site where much dirt could be expected to accumulate. However, it was considered representative of many parts of the tourist cave system.

The experimental site was a sloping (approximately 10-15°) finely crystalline flowstone slab which had been truncated by the cutting of the tourist pathway. Visual examination under ultra-violet light revealed the presence of very many fibres derived from fabric which had been either bleached or washed with optical brighteners (which fluoresce at ultra-violet wavelengths).

The use of the microerosion meter

Installation of the two MEM sites required the drilling of holes in the flowstone. Dust contamination was reduced by removing dust continuously during drilling with a vacuum cleaner and by shielding the measurement site with cardboard. Closed examination of the MEM sites before and after drilling and installation of studs showed that calcite dust contamination of the two MEM sites was minimal - that is, drilling dust was not greatly increasing the height of the surface and thus not exaggerating lowering following washing.

Four sets of surface lowering measurements were made on MEM site 1; one before washing and the others after each of three following washes. Error on part of the experimenters resulted in the second washing of site 2 before MEM determinations were made and thus the sites cannot be directly compared for changers after the first wash although comparisons can be made after washes 2, 3 and 4. As the two populations of observations are not significantly different at p=0.05 they have been pooled and the discussion below relates to these pooled data.

The results of the MEM measurements are set out in Table 1.

Table 1. Cumulative surface lowering following high pressure washing in Imperial Cave, Jenolan Caves, NSW.

 Cumulative loss after each wash (millimetres)
 Wash 1Wash 2Wash 3Wash 4
standard deviation0.0150.0280.0440.047
number of observations20202020
coefficient of variation74%95%82%70%
mean 0.0250.0500.078
standard deviation 0.0370.0460.066
number of observations 202020
coefficient of variation 145%92%84%
BOTH SITES (after four washes)
mean 0.0280.0510.072
standard deviation 0.0330.0450.057
number of observations 404040
coefficient of variation 119%87%79%

SEM sampling strategy

Three sub-sites were selected along the lower edge of the flowstone slab. Samples were easily chipped from here without causing major damage to the flowstone surface or affecting MEM readings. Care was taken to ensure that chippings or powder generated during sampling did not contaminate the surface of the sample. Specimens were removed before washing and immediately after each subsequent wash.

During each wash the run-off was collected and then filtered through 0.47μm filters to remove suspended sediment. The samples were allowed to dry at air temperature before preparation for analysis.

Both flowstone and sediment samples were mounted on 1cm aluminium stubs and coated with gold for study in a Jeol 6400 Scanning Electron Microscope.

The supernatant wash water was analysed by Dr Julia James of the University of Sydney. The wash water is derived from the river in Imperial Cave and is at or near saturation with respect to carbonate.

The washing process

A high pressure washer delivering approximately 5500-7000kPa (800-100psi) was used by Jenolan Caves staff familiar with its use in actual washing projects. The actual operating pressure was somewhat less than 5500kPa as operating problems were encountered. Bonwick and Ellis (1985) state that:

"From the Jenolan experience there would seem to be little or no advantage in pressures higher than 750psi (5200kPa). In fact any higher pressures could result in serious damage to calcite surfaces."

Care was taken to reproduce 'normal' washing practice but this is necessarily subjective and broad scale use may be more or less damaging than the washing we employed.


Results from the scanning electron microscopy

On the unwashed flowstone surface individual calcite crystals are mostly obscured by an amorphous layer of clay material. The presence of clay is confirmed by electron microprobe analysis which detected the elements iron, potassium, aluminium, magnesium, sodium and calcium. Most of this probably originates from atmospheric dust but may have also come from cave sediments or from soil washed down into the cave through joints from the surface. The clays coat individual crystal faces and are concentrated in the interstices between them.

Other contaminants on the unwashed surface visible in the micrographs include hair and lint fragments originating from cave visitors. These fragments are found resting on the clay layer, partially or wholly coated by clay and embedded within the calcite itself. Surprisingly skin fragments were not observed.

One wash is sufficient to clean off most of the clay material, hair and lint fragments. The result of this is to expose individual calcite crystals and to clear out the interstices. Small amounts of clay material still adhere to the larger faces of the calcite rhombs.

The second wash effects the further removal of clay material and exposure of calcite crystals. But mechanical damage to the surface of the flowstone is also evident. Firstly whole individual calcite crystals are removed and, secondly, fragments of crystals are removed as a result of mechanical fracturing.

Subsequent washes have much the same effect. Most of the clay material has now been removed and all the hair and lint has gone. The effect now is mainly the mechanical removal on individual crystals and crystal fragments. The depth of crystals removed is about 0.1mm (compared with 0.07mm determined by the MEM) to form micropits of approximately 0.7 x 0.5mm. It appears that the surface layer of crystals is removed as these pits coalesce following the removal of the walls between pits producing a relatively uniform, crystalline surface.

In all cases the size distribution of the wash sediment was bimodal. It consisted of larger masses of the order of several square millimetres and small fragments of the order of several 100ths of a square millimetre. Sediment from the first wash also contains fragments of hair and lint, most of which have clay particles adhering. The larger masses consist of agglomerations of calcite crystals and clay material. The finer sediment is composed mainly of clay material, but also of some calcite crystals which exhibit fracture patterns.

In subsequent washes very little, if any hair or lint is evident having been mostly removed in the first wash. A few hair fragments were found in the material from the third wash. It is possible that these were washed in from an adjacent area. Alternatively, and more probably, these may have been cemented within the flowstone and removed by mechanical removal of calcite crystals. The crystals of washes 2, 3 and 4 were increasingly "clean" in that no clay particles are adhering to them. Instead, individual crystals, crystal fragments and groups of crystals are removed mechanically and exhibit conchoidal and stepped fracture patterns.

Results from the microerosion meter

As can be seen from Table 1 the variability of lowering of individual MEM points is high but this is to be expected from the technique. However, increasing numbers of observations and, probably, the removal of loose material from the surface, show a clear trend with about the same amount of lowering with each wash (about 0.018mm per wash; Figure 1). This is a similar rate to that estimated from the SEM observations.

Water chemistry

As mentioned above the wash water is drawn from the Imperial Cave streamway and is at or near saturation with respect to dissolved carbonate. This is good washing practice and the use of saturated cave waters should always be used if feasible.

Analyses were carried out for a range of anions and cations and no real differences were discernible as would be expected from the saturation of the wash water and the very limited contact time between water and flowstone. If anything the level of total dissolved solids declined by about 20% (from 225ppm) through the experiment from wash one to five perhaps reflecting lesser amounts of finer material available for solution in subsequent washes. This may also relate to changes in wash water quality before washing.


Both the SEM and MEM results demonstrate that high pressure cleaning changes flowstone surfaces. The SEM shows the plucking out of calcite blocks some 0.1 x 0.07 x 0.05mm. Pits of this size will not be well sampled by the MEM because of the large diameter of the hemispherical probe (relative to the pits). However, the rates of surface lowering are consistent between approaches and the value of using two complementary techniques are evident.

A lowering of less than one tenth of a millimetre over four washes may seem trivial. However, it must be remembered that the first wash produces a visually cleaner surface by removing about one quarter of the total lowering and thus the next three washes are doing two to three times as much work even if this cannot be seen directly. It could be argued that four washing sequences one after another do not reflect current cave cleaning practice where washing is carried out at infrequent intervals, often several years apart.

However, in practice some areas are washed repeatedly in one cleaning sequence and this raises some questions. For example, when a roughened surface is created do the calcite crystals grow in the same way on the etched plane? Do roughened surfaces collect and hold more debris? These questions deserve further thought. It could be expected that there would be some recovery of the surface in these cases and this would obviously be so in areas where calcite is actively being deposited.

It can be concluded from the data presented that high pressure washing does have an impact but that this is relatively minor at or below about 5500kPa (800psi). Higher pressures should be avoided. It would appear desirable to keep the number of passes of the nozzle as few as possible and to stretch out the time between cleaning although the danger of having the debris cemented with the attendant problems of a more intensive job being required. The possibility that the cemented debris mat may either strengthen or weaken the surface layer must also be considered. Changes could mean either more force is required or that more care is needed.

Figure 1: Jenolan Washing Experiment


We thank the Jenolan Caves Reserve Trust who made the site, their equipment and staff available for this work. In particular Ernst Holland and Nigel Scanlan provided both a labour force and background information on past and present washing practice at Jenolan.


Aley T, Aley, C and Rhodes, R, 1985 Control of exotic plant growth in Carlsbad Caverns, New Mexico; Proc 1984 National Cave Management Symp., Missouri Speleology 25(14)159-171

Amoroso G G and Fassina V 1983 Stone decay and conservation, Elsevier, Amsterdam, 453p

Bonwick J and Ellis R 1985 New Caves for Old: Cleaning, Restoration and Redevelopment of Show Caves, Cave Management in Australasia VI Waitomo; pp134-153

Krinsley D H and Doornkamp J C 1973 Atlas of quartz sand surface textures; Cambridge University Press

Lewin S Z and Charola A E 1978 Scanning Electron Microscopy in the diagnosis of "diseased" stone; Scanning Electron Microscopy, 1:696-703

McGreevy J 1982 Frost and salt weathenng fuither experimental results; Earth Surface Processes and Landforms, 7:475-488

Newbould R L, 1976 Steam Cleaning of the Orient Cave, Jenolan Caves, NSW; Cave Management in Australia I, pp87-9O

Spate A P, Jennings J N, Smith D I and Greenaway MA 1885 The Micro Erosion Meter Use and Limitations, Earth Surface Processes and Landforms 10:427-440

Trudgill ST, Viles H A, Inkpen RJ and Cooke R U 1989 Remeasurement of weathering rates, St Pauls Cathedral, London, Earth Surface Processes and Landforms, 14:175-196

Viles, H A 1990 The early stages of building decay in an urban environment, Atmospheric Environment, 24A(1 )229-232