Chapter 2

PRODUCT THEFT OF GOLD

EXISTING ESTIMATES BY THE INDUSTRY

The only existing estimate of the extent of the theft of gold from mines and refineries in South Africa is the figure provided by the Chamber of Mines of 30 tons valued at approximately R1.58 billion that is estimated to have been stolen during 1996.

This figure has been mentioned on a number of occasions by the industry. An attempt to establish the origin of this figure revealed the following facts.

During a visit to a European gold refinery, an official from the South African gold industry was asked what the annual figure was for the theft of gold in South Africa. According to the official concerned, he mentioned the figure of 11 tons for the year 1996. The European refinery representative apparently dismissed this figure claiming that the refinery itself had bought approximately 25 tons of South African gold from unofficial sources during 1996. During a trip to another refinery in Europe, the South African visitor was informed that the refinery in question had bought approximately 7 tons of South African gold from unofficial sources during 1996. The information from both refineries was given to the South African during casual discussions and on an informal basis. Neither of the two refinery representatives was apparently prepared to divulge more information and there was no follow-up to obtain confirmation of these figures from the refineries concerned.

On the basis of this information, the South African official concluded that a conservative estimate of gold stolen in South Africa during 1996 was 30 tons for the year 1996, although he believed it to be closer to 50 or 60 tons. The information given to the South African visitor could not be verified, nor is there any basis on which to suggest that the information supplied by the European refinery representatives was false. The information is unconfirmed and uncorroborated, however, and was published in the Annual Report of the Chamber without any closer scrutiny of the claims having been undertaken. In order to have a degree of credibility, estimates should be based on projections or calculations derived from a factual basis that can be verified.

Even though there are no facts available on the basis of which the figure of 30 tons can be confirmed or refuted, the assessments made in this study were not influenced by it. The point of departure for the study was therefore that the extent of theft from gold mines and refineries in South Africa was unknown.

POLICE STATISTICS

The Diamond and Gold Branch of the South African Police Service (SAPS) rendered every assistance possible in support of this study. Its national statistics relating to the recovery of unwrought gold and arrests, among others, are set out in figure 1. The information was supplied by the Pretoria headquarters of the Diamond and Gold Branch based, in turn, on statistics supplied by local Diamond and Gold Branch offices in the provinces.

The mass of unwrought gold recovered nationally by the Diamond and Gold Branch can only be an indicator, albeit a very inadequate one, of the prevalence of theft of gold from mines. These statistics are inevitably based on estimates. Recovered unwrought gold consists of different concentrations, ranging from the gold-bearing ore which thieves carry out of gold mines to well refined gold. To attach an accurate value to such recoveries is difficult. The gold content of a bag of gold-bearing material can only be established once such ore has passed through the processing of a gold mine. Even then such ore is not refined separately from the regular ore recovered by the mine. In practice, the police return unwrought gold that they confiscate to the mine from which, in their assessment, such unwrought gold or gold-bearing material originates. The mine in question thereafter refines the gold bearing material and informs the police of the mass and value of extracted gold. The prime indicator of the extent of unwrought gold seized or found by the police is therefore the value that mines place on the gold-bearing material received from the police.

The general approach of police headquarters in such cases is to attach a value of R50 to a gram of refined gold. When a mine therefore informs the police that the value of the stolen gold-bearing material handed in by them for processing was R100 000, the police will record the mass of this gold as 2 000 grams. In order to compute annual national statistics, the police add up the value that mines have attached to the unwrought gold and gold-bearing material that were supplied to them for processing and divide the sum by 50 in order to arrive at the estimated mass of the gold. This was also the method applied in arriving at the mass of unwrought gold recovered as set out in figure 1 below. The valuation of R50 per gram of gold is in fact conservative if account is taken of the fact that the average gold price on the London gold market for the years 1994 to 1998 was R53.39 per gram.11

This method of computing the mass of stolen gold recovered by the police has its shortcomings. Precisely how do mines deal with stolen gold that the police return to them? Do all mines follow the same procedures when processing and determining the value of such unwrought gold? Do all Diamond and Gold branches in the country follow a uniform approach in this regard? It is well known that some top dealers in illicit gold sometimes add gold from Kruger coins to gold bought on the illegal market when smelting the gold for export. This happened in the widely publicised Chemfix case. Should the police confiscate such a consignment, or should the South African syndicate leader sell such gold to a European refinery, the mass of this gold will clearly not provide an accurate reflection of how much gold was stolen from mines and refineries in South Africa. The question of how to reflect the mass of detected gold stolen from mines and refineries is therefore an aspect that requires further examination by both the police and the gold mining sector.

It has happened, for example, that the police and mines have agreed that the police would not publish information relating to the mass or value of confiscated gold-bearing material returned to the mines. Such information would then clearly not form part of police statistics and would only be available to the gold mines. This was the case in the Welkom area with gold-bearing material that was confiscated during raids on the well-known G Hostel in Thabong township outside Welkom. Clearly, such ad hoc arrangements between the police and the mining industry make it very difficult to obtain accurate statistics relating to the extent of the theft of gold from mines.

Even if the national statistics supplied by the police were an accurate reflection of the mass of stolen gold recovered by them, it is clear that what they recover can only be a fraction of the actual volumes that are stolen on an annual basis. The regional case study referred to below, and other indicators to which reference is made in this study, confirm this assertion.

The mass of gold in the summary that follows was arrived at by attaching a value of R50 per gram, based on a division of the total value of the gold recovered by 50. If the actual average gold price per gram over the five-year period — R53.39 per gram — had been used as a measure, the total mass of the gold recovered by the police over this period would have been 1 297 kilograms instead of the 1 385 kilograms reflected above.

Police statistics (as illustrated in figures 2, 3 and 4) show two different trends. While the number of arrests for unlawfully purchasing gold, for being in unlawful possession of gold and for the theft of gold has shown a gradual decline since the peak reached in 1995, the figures relating to the mass and value of gold recovered by the police show a marked increase since 1996. It suggests that the quantity of gold-bearing material that is being recovered or confiscated by the police is higher per person arrested than in the past. This could also indicate a greater sophistication among gold smugglers and thieves enabling them to deal in larger quantities of gold. This would mean that they are more difficult to arrest than was the case three or four years ago. The decrease in the number of arrests, however, could also have been influenced by a general shortage of personnel and resources in the various Diamond and Gold Branches in the country.

The statistics provided by producers of more than 60% of South Africa’s total gold production have been presented in this report to reflect a view of the entire gold-mining industry. It was assumed that the statistics obtained from these producers are a reasonable indication of what occurs at those mines that produce the remaining 40% of the country’s gold. Those mining companies that assisted with information applied different criteria in producing their statistics. These are all problems and limitations that should be borne in mind when estimates of the extent of product theft of gold are made in this report.

Statistics supplied by mines indicate that approximately 483 733 kilograms gold-bearing material, that was stolen or suspected to be stolen, were recovered by mines during the five years from 1994 to 1998 (figure 7). When the 22 810 kilograms of gold and gold-bearing material recovered by the police (figure 1) are added, the total comes to approximately 506 543 kilograms of gold-bearing material recovered in South Africa over the five-year period.

In this report, very little significance is attached to the mass of gold-bearing material recovered, as such figures bear no relation to the actual gold content of the material. It is the actual gold content and the value that will be concentrated on.

Figure 7 shows values that were computed (and adjusted) on the basis of the statistics supplied by mines. They suggest that gold mines recovered unlawfully removed gold-bearing material worth approximately R87 939 160 during the five-year period from 1994 to 1998. The value recovered by the police was R69 258 024.

The total value of gold recovered by both the SAPS and the mines over the five-year period is summarised in figure 7 and graphically illustrated in figures 8 and 9. The figures show an overall decline during 1997 and 1998 in the recovery of unlawfully removed gold-bearing material. Marked differences, however, appear in the recovery trends between the gold mines and the Diamond and Gold Branch of the police. Figure 8 illustrates that, particularly from 1997 onwards, the police recorded significant increases in their recovery rates, while the value of gold-bearing material recovered by mines went into a strong decline, reaching a five-year low in 1998. The reasons for these contradictory trends are not clear.

One explanation for the decline in the value of unlawfully removed gold-bearing material recovered by mines could be the more effective security measures that many mines have introduced. If this is the case, how can the increased recovery rate of the police be explained? It may be due to improved policing methods, or to the apparent increase in the volume of stolen gold found on persons arrested. It must be assumed that the gold retrieved by the police largely originated from gold-bearing material that passed undetected through mine security systems.

Of interest is the indication, based on mine statistics, that by far the largest volumes of gold-bearing material is unlawfully removed from gold plants rather than from shafts or elsewhere. If the statistics that were provided by a significant roleplayer are adjusted to apply to the gold-mining industry as a whole, it would appear that, in 1998, approximately 71% of the gold-bearing material recovered by mines originated from gold plants, about 23% from shafts, and the remaining 6% from unknown sources. The high rate of removal of gold-bearing material from plants is a clear indication that this part of mine operations still constitutes the highest risk area despite improved security.

GOLD REFINERIES

Refineries have a less difficult task than mines when assessing the mass and value of gold stolen from them. A careful record is kept of all gold delivered to the refinery, as well as of the gold, once refined, that leaves the premises. Any difference between the input and the output of the refinery could be ascribed to two factors only: process loss and theft.

Refineries, as has been the case with some mines, have also been reluctant to reveal the extent of the theft of gold in their custody. Rand Refinery Limited, by far the largest refinery in the country, made some statistics available that are reflected in figure 10. The refinery has determined that up to 0.02% of the total mass could be ascribed to process loss. However, it is accepted that some percentage of this portion is attributable to theft.

According to the refinery, there have been significant improvements in the provision of more effective security. In the past, approximately 99.98% of the gold delivered was recovered from the refinery process. The loss of the balance (0.02%) was ascribed to theft. If this formula is to be applied to 1998, the mass of gold stolen would have been approximately 98.28 kilogram, valued at R4 914 000. The refinery is of the view that additional security measures, as well as the improved recovery methods that have been introduced during the past few years have by now produced an almost 100% recovery rate. The information that was made available indicates that, between 1 January 1994 and 23 September 1999, the refinery recovered a total of 8 779 kilograms of stolen unwrought gold, valued at R438 950.

A reasonable estimate cannot be made of the losses resulting from undetected theft at the refinery. If undetected thefts had occurred, it would have been virtually impossible to determine due to process losses. The gold stolen and recovered from the refinery can therefore be considered insignificant as a contribution to the overall volumes of gold that is smuggled in South Africa.

TOTAL DETECTED PRODUCT THEFT OF GOLD

An assessment of the total product theft of gold that relies on statistics provided by the police, mines and the refinery has obvious limitations. Figures only reflect those thefts that were detected and not the ones that remain undetected. Figure 11 summarises the extent of detected gold thefts over the past five years in respect of which official statistics are available, albeit in an imperfect form. The figures are confined to stolen gold recovered.

On the basis of the values provided in Figure 11, the mass of the detected theft of gold per year can be calculated by dividing the annual value of detected gold thefts by the average gold price per gram during a particular year. The results are provided in figure 12.

Even though the gold referred to in Figure 12 was stolen, or suspected to be stolen, it was in fact recovered by mine security or by the SAPS and returned to the mines or plants. The amounts reflected in Figure 11 can therefore not be regarded as constituting a loss to the gold production figures of the gold-mining industry. Significant expenses would have been incurred both by the state and the mining industry in recovering the stolen gold, but it will have been reported as part of the annual gold production of the industry. The fact that the gold was returned, however, does not negate the fact that the returned gold is correctly referred to as ‘stolen’.

The value of the gold known to have been stolen from mines between 1994 and 1998 amounted to an average of approximately R32 million per annum, involving an average of about 603 kilograms of stolen gold per annum. However, these figures do not reflect the actual (undetected) theft of gold.

There are many limitations that need to be borne in mind when considering the above statistics. For example:

Some of the gold-bearing material recovered by the police is occasionally returned to the relevant mines without being reflected in police statistics.



Some of the unwrought gold recovered by the police contains elements of gold that was not necessarily stolen from mines. Licensed gold dealers and pawnbrokers sometimes mix stolen gold that they have purchased with gold legitimately obtained, such as gold jewellery, and then forward it to refineries for refining.12



Police statistics reflect volumes of gold-bearing material that are calculated by attaching a value of R50 per gram of gold that was extracted from the confiscated gold-bearing material.



The statistics relating to the theft of gold-bearing material from mines are based on the actual statistics provided by producers responsible for more than 60% of South Africa’s total gold production. These statistics were used to calculate figures that would apply to the entire gold production in the country.



The statistics relating to gold refineries are based on information from one refinery only. However, the extent of theft from refineries appears to be insignificant and unlikely to make any difference to the total figures.

The challenge remains to explore whether enough facts and circumstantial information are available to justify an estimate of undetected gold thefts.

AN ASSESSMENT OF UNDETECTED THEFT OF GOLD 

The information supplied by mines suggests that the value of gold-bearing material known to have been stolen has declined between 1994 and 1998 (figure 7). The Diamond and Gold Branch statistics indicate that the number of arrests relating to the theft of gold have remained more or less constant during the same period, with a slight decrease experienced during 1998 (figure 2). The only statistical indicators that could suggest that the incidents of theft from mines might have increased during the past few years, are the increased recovery rates attained by the police, both with regard to the mass of gold-bearing material and the value of gold recovered (figures 3 and 4). To what extent this increased rate of recovery is influenced by more effective police investigative measures, or to what extent it is due to increased (but undetected) thefts from mines is difficult to say. What is clear though, is that it is not possible to conclude that the gold-mining industry is experiencing a serious increase in the theft of gold-bearing material from mines from the statistics provided by mines and the police.

Yet, in the clear and almost unanimous view of gold mine security personnel and members of the Diamond and Gold Branches who were interviewed in different parts of the country, the theft of gold has increased significantly. They all share the belief that both mine and police statistics reflect merely the tip of the iceberg. Their conclusions are based on a wide range of factors that warrant some examination.

INCREASE IN THE NUMBER OF 'SMALL-TIME' GOLD SMUGGLERS

Both the police and mine security personnel are of the view that there has been a significant increase in the number of small-time operators involved in theft from mines. This is ascribed to the general state of the economy with its high unemployment rate, but also to the retrenchment of thousands of mineworkers in the past decade. Many retrenched mine workers, who have gained considerable experience and knowledge about the mining process, soon return from their homes in far-flung areas or from neighbouring states, to try and make a living near the mines with activities best known to them: the mining of and dealing in gold. They link up with local illegal smelting houses or smuggling rings and set about obtaining gold in whatever form, and from any location possible. Some are known to be digging tunnels into disused mine dumps with the hope of reclaiming some gold.

This category of operators provide the runners and those who, in the Welkom area and elsewhere, find their way into shafts at night to remove gold-bearing material. According to mine security personnel, such night mining did not take place four or five years ago. They regard it as a new trend. Retrenched workers tend to know many of the regular mine workers who work at the mine in question, and therefore have no problem in borrowing their security cards to gain entry into the mine. They may not individually steal large quantities of gold-bearing material, but collectively their activities must account for significant volumes of gold stolen.

If their numbers have increased over the past few years, which by all accounts seems likely, then this is not reflected in police statistics or in the adjusted mine arrest rates as shown in figure 13.

A drop in the number of arrests by both the police and mines, during a period when there seems to have been a significant increase in the number of small-time gold thieves, could actually explain why their numbers have increased. It seems clear that they are operating in a relatively risk-free environment with a very slim chance of being arrested. In the absence of an increase in arrests, their contribution towards the theft of gold from mines remains unrecorded and not reflected in police or mine statistics.

It is therefore reasonable to deduct that far more gold is likely to be being stolen from mines by small illegal miners and runners than official statistics suggest. It is, however, an impossible task to determine accurately the quantities that are involved. If there was more success in arresting the main players, such as the sophisticated syndicate leaders in Gauteng, Durban and elsewhere, a start could be made in accurately estimating the extent of undetected gold thefts. The few grams of gold that the illegal miner may recover per week, when multiplied by the number of those active in illicit mining, are likely to add up to significant volumes. These small quantities are sold and smuggled through the various hierarchies of syndicates until the few large buyers who operate in South Africa buy them up. When arrests and confiscations take place at this level, it will be less difficult to draw inferences about the levels of undetected theft of gold from mines.

THE ROLE OF MIDDLE LEVEL AND TOP SYNDICATES DEALING IN GOLD

While the retrenched mine worker, who returns to the mine to retrieve gold-bearing material illegally, is situated at the bottom of the ladder of hierarchies in the criminal world, the more sinister role of the top structures that are involved is seldom highlighted. These are the sophisticated and well-organised criminal groups to whom the stolen gold is channelled and who then dispose of it internationally or locally. Because of the lack of success by the police in prosecuting these groups, little is known about them. Among members of the Diamond and Gold Branch, however, there is a fairly accurate notion about the identity of individuals who are involved. They know the names and addresses of the key roleplayers, but always add that those individuals are difficult to prosecute because they tend not to get personally involved. Many of them have been operating for many years without being prosecuted. Other than the well-known Chemfix case, there have been very few police successes against the larger operators involved in the theft of gold. In the Chemfix case, 7.7 tons of gold obtained over a period of about one year, were involved. A portion of the 7.7 tons, however, did originate from Kruger coins.

In one of the few other successful operations involving a sophisticated organised group, the Randfontein Diamond and Gold Branch, acting on a tip-off, seized approximately 20.3 kilograms of gold worth approximately R1 million from a private jet that was about to depart to Geneva from Lanseria airport in 1995. There were indications that the plane was to travel via the Rand Airport to take on another load of smuggled gold. The pilot, who was apparently paid an amount of R10 000 as a courier reward, was arrested and so was the person who supplied the gold. The supplier of the gold was eventually convicted in 1999 and sentenced to a fine of R15 000 and three years imprisonment, suspended for five years — a sentence that can only reinforce the view of top gold syndicate members that their illegal ventures carry low risks. The fingerprinting of the confiscated gold indicated that its source was both the West Rand and the Free State gold fields, a good example of how small amounts of gold recovered by the individual illegal miner are eventually bought up by the top syndicates with international links.

Within the gold-mining and the Diamond and Gold Branch fraternity, different views exist about the number of top crime syndicates that are actively dealing in stolen gold. A senior person in the industry claims that there are six such syndicates. A top person in charge of security for a very large gold-producing company is of the view that ten to eleven such syndicates are active. The police are not certain about the number of top syndicates involved. They admit that they mainly tend to catch the ‘small boys’ or runners and occasionally members of larger middle level syndicates. They are not sure about the end market for the gold except that some of it leaves regularly for Hong Kong and Bangkok in the form of machine parts. Others believe that most of the stolen gold ends up on the internal market.

Because so little is known about the extent to which the ‘top boys’ deal in stolen gold, local Diamond and Gold Branches regard the larger buyers or syndicates operating in their area as falling into the category of ‘larger middle level gold buyers or syndicates’. They are not regarded as part of the few top syndicates or buyers in the country. Some members of the Carletonville Branch, for example, have identified six big syndicates within their area of jurisdiction. Within the area of jurisdiction of the Randfontein Diamond and Gold Branch, there are at least four large syndicates that are large buyers of gold. According to a senior person involved with mine security, there are about 22 big buyers of illegal gold in the West Rand area, with 12 to 13 of them operating syndicates. He also referred to 85 illegal smelting houses that operate in the area.

The various Diamond and Gold Branches are constantly engaged in covert operations against syndicates within their area of jurisdiction. This often occurs with the involvement of mine security staff. Some of the local gold-buying syndicates that they investigated appear to have a considerable turnover of stolen gold. On the West Rand, for example, the police recently searched the house of the leader of such a syndicate and confiscated 8.1 kilograms of gold nuggets They estimate this to have been one week’s stock. Such a syndicate can probably be regarded as a middle level syndicate on the ladder of hierarchies. In another operation, police bought 43 kilograms of unwrought gold, worth R1.2 million, from a syndicate over a six-month period. A total of 81 arrests were made. This category of syndicate, even though it is of local importance, does not fall into the category of the top syndicates with international connections. In another operation in Johannesburg during 1997 police bought 63.8 kilograms of unwrought gold valued at approximately R2.5 million over an eight-month period.

Welkom Diamond and Gold Branch members have identified 25 syndicates of various sizes in their area. They maintain that most of the gold stolen in their area is eventually moved from Welkom to Gauteng and to a specific syndicate that they have identified. During an eight-month long undercover operation in 1999, the Welkom Diamond and Gold Branch bought 90 kilograms of stolen gold worth approximately R3.5 million from various small-time gold thieves who had stolen it from local mines. Their conservative estimate is that approximately five kilograms of unwrought gold leaves the G Hostel complex on a daily basis. That would add up to 1.8 metric tons per annum.

The head office of the Diamond and Gold Branch provided the most reliable information relating to the number of larger middle level gold dealers or syndicates operating in South Africa. After an exercise that also involved the various branches in the regions, the police concluded that there were approximately 36 larger middle level gold dealers or syndicates operating throughout the country (figure 14).

ESTIMATES: UNDETECTED THEFT OF GOLD

As mentioned above, very little is known about the top gold-buying and selling syndicates. When local branch members were asked why they did not investigate the well-known wealthy buyers/sellers in Gauteng, their response was that those syndicates fall into the area of jurisdiction of another branch and that a file or docket appeared to have been opened by another branch in respect of those syndicates. As a result, few of the top syndicates have been successfully investigated and the extent to which they constitute a threat is difficult to assess.

To rely on the few successful exposures of the top syndicates, or on anecdotal information, with the aim to estimate the extent of the undetected theft of gold would not lead to credible projections. The information that is available about larger middle level syndicates or gold dealers provides a more reliable base from which to make estimates. This is the only avenue through which it might be possible to arrive at an estimate of undetected gold thefts. However, the following three assessments were made without being confined ourselves to larger middle level gold dealers or syndicates.

Estimate 1 (small syndicates/buyers): If police on the West Rand, who searched the house of a local syndicate leader and found 8.1 kilograms of gold, are correct in their assessment that this represents only one week’s takings, then it follows that over one year such a seller/syndicate would deal with about 421 kilograms of stolen gold. On the assumption that there are at least 50 syndicate/sellers with a similar capacity active in South Africa, such syndicates would together account for a turnover of about 21 tons of stolen gold. The assumption that at least 50 similar sized syndicates are active in the country is by no means far-fetched if the following is taken into account. According to a senior mine security official, there are 22 big syndicates/buyers of stolen gold in the West Rand area alone. The Welkom Diamond and Gold Branch has identified 25 syndicates within its area of jurisdiction that buy stolen gold. Not all gold-mining areas or Diamond and Gold Branches were canvassed for information about the number of syndicates operating in their areas. Even without having undertaken such an exercise, it can confidently be assumed that there are well in excess of 50 lower level small-time syndicates/buyers in South Africa that deal with similar volumes of stolen gold.

Estimate 1 suggests that the undetected theft of gold from mines and shafts could be approximately 21 tons per year.

Estimate 2 (larger middle level syndicates/buyers): During one of their recent raids on G Hostel, the Welkom Diamond and Gold Branch seized a record book kept by one of the sellers of gold in the hostel for the purpose of recording the particulars of his transactions. The notes reflected that the seller had received R747 664 in 66 days and that, on average, he sold 2 882 grams of unwrought gold per day. During one week, this person would have sold, on average, about 20 kilograms of gold and during one year, about one ton. This was just one of many sellers in the hostel. Many of the other sellers in the Welkom area operate from outside G Hostel.

Estimate 2 suggests that the undetected theft of gold from mines and shafts could be approximately 35 tons per year.

Although this seller by no means falls into the category of the top sellers/syndicates in the country, he seems to have a more significant turnover than the 8.1 kilograms per week of the syndicate leader from the West Rand example referred to above. It would be more appropriate to regard this dealer as falling into the ‘middle level’ category of operator or syndicate. The number of sellers/syndicates at this level of operation is likely to be smaller than the estimated 50 referred to in the previous example. Therefore, on the basis that there are at least 35 buyers/syndicates in South Africa of a similar capacity as the one from G Hostel, a total of 35 tons of stolen gold would have changed hands due to their activities in one year. Some members of the police and mine security regard the estimate of about 36 syndicates/buyers with a similar capacity operating in South Africa as conservative. The view of a senior mine security officer that there are 22 big syndicates/sellers in the West Rand area alone, can also not be ignored. However, the police are of the view that the figure of 22 includes several smaller syndicates that operate on a smaller scale.

Estimate 3 (top national syndicates/buyers): In the Chemfix case, a mass of 7.7 tons of gold were involved. Not all of it originated from illicitly obtained gold, as a portion of the 7.7 tons was made up of gold from Kruger coins. Ignoring this aspect, and on the assumption that there are at least six top sophisticated syndicates active in the country that deal in similar volumes of stolen gold, it could be deduced that these six syndicates could have a joint turnover of about 42 tons of gold per annum. Although this projection of 42 tons of gold being stolen in South Africa per annum may not be far-fetched, it is based on insufficient factual information to give it credibility. No reliable information could be obtained about the number of top gold-dealing syndicates in the country.

Estimate 3 suggests that the undetected theft of gold from mines and shafts could be approximately 42 tons per year.

While the 42 tons may well be realistic, estimate 3 is unlikely to have sufficient credibility in the absence of more information on which such an estimate can be verified.

The theft of 21 tons and 35 tons as suggested by estimates 1 and 2, respectively, seems more realistic. However, these remain estimates that are based on insufficient evidence to justify elevating them to the level of statistical fact. It is not known, for example, what the exact degree of purity of unwrought gold was that the dealer sold over the 66-day period. Despite these limitations, and taking into account all the information that was obtained during the process of undertaking this study, the estimate of 35 tons of undetected gold stolen from mines and plants per year is not unrealistic. It may well have been more, but there would be too much speculation involved in attempting to explain such volumes. The capacity of the larger middle level syndicates and buyers in all likelihood have expanded since 1994. However, it is not unreasonable to assume that the estimate of 35 tons could have equally applied to any one of the years between 1994 and 1998.

It is therefore estimated that the undetected theft of gold from mines during the five years from 1994 to 1998 amounted to an average of 35 tons per year.

In order to arrive at the average mass of detected plus undetected theft of gold per year during the five years from 1994 to 1998, the 603 kilograms of gold recovered on average per year by the police and mines have to be added. The mass of gold stolen from mines between 1994 to 1998 can therefore be said to be an average of 35.6 tons per year.

It is therefore estimated that an average of 35.6 tons of gold were stolen per year from mines in South Africa during the five years from 1994 to 1998. Of this total, an average of 603 kilograms per year were recovered by the police and by mines.