GB2407172A

GB2407172A - Counting cash by weighing

Info

Publication number: GB2407172A
Application number: GB0324162A
Authority: GB
Inventors: Christopher Lare
Original assignee: Tellermate Group Ltd; Tellermate PLC; Tellermate Ltd
Current assignee: Tellermate Group Ltd
Priority date: 2003-10-15
Filing date: 2003-10-15
Publication date: 2005-04-20
Also published as: WO2005038732A8; GB0324162D0; WO2005038732A1

Abstract

A method of counting cash comprising repeatedly weighing a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; after each weighing, determining the number of items in the group; comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings; as a result of the comparison, determining whether items have been added or removed; choosing a respective counting method in accordance with whether items have been added or removed; and counting in accordance with the method chosen.

Description

Imorovements in cash counting This invention concerns improvements in cash

counting, especially to the accuracy of cash counting machines that employ weighing techniques in determining cash values.

In retail sectors it is a longstanding problem that the value of cash held within cash registers is not immediately obvious and determining this value to accurately requires a time-consuming manual count operation. The value of cash held within cash registers is important to know for management purposes, for example, to determine when a cash drawer is full, or nearly full, and an amount of cash (known as a float) needs to be removed from that drawer. This action is commonly referred to as 'skimming' or'lifting' cash from a cash drawer. Cash values are also needed to determine cash levels within a cash register drawer during the day so that cashflow information may be obtained and fraudulent practices may be spotted. In all cases, it is desirable that accurate cash value data is obtained frequently throughout the working day or shift, that this is done with as little intrusion as possible to the operation go of the cash register and that as little operator intervention is required as possible.

An example of an 'intelligent' cash register is disclosed in European patent application EP 0724242 to Tellermate Plc., which describes a cash register comprising weighing means arranged to take weight readings from the cash compartments of the cash register. The weight readings of each compartment may then be converted to a cash value by a processing unit so that the total cash value contained in the register may be obtained without manual intervention.

It will be appreciated that the term cash used herewith does not merely refer to banknotes and coins but also to any type of accepted tender or cash item, for example prorrissory notes, vouchers, tokens and the like. For ease of reference, the foregoing items based upon absorbent substrates like paper or card, shall be referred to simply as 'banknotes', and the remainder shall be referred to simply as 'coins' in the rest of this specification.

With intelligent cash registers and all cash counting machines employing 6 weight techniques, a fundamental principle to their accurate operation lies in their reliance upon a recognised standard weight for each type of banknote or coin being weighed. This standard weight is usually determined under normal ambient conditions and kept in a memory store for subsequent use during a count procedure.

However, a problem faced by all counting machines employing weighing techniques is how to accommodate frayed, torn or worn banknotes and coins, as these 'distressed' banknotes and coins may weigh appreciably different from the pre-determined standard weight. In known counting machines 16 utilising weighing techniques, it is common for a simple rounding procedure to be used to round either up or down to the nearest whole banknote or coin.

Accordingly, for example, a 2% weight differential between the actual weight of a cash item and the pre-determined standard weight of that cash item is enough to register a miscount be a whole banknote or coin with as little as 26 go banknotes or coins being weighed. This drawback of this type of simple but prevalent rounding procedure is evident.

To provide for this, cash counting machines employing weighing techniques can use several methods to improve the accuracy of their count procedures, 26 and a non-exhaustive sample of these are detailed below for example. These methods, or algorithms, work in a satisfactory manner although some algorithms are more appropriate to certain types of cash situation. In all the examples below reference is only made to counting banknotes for conciseness; however, it is appreciated that the same procedures will be equally applicable to all forms of cash items. Furthermore, all counting machines employing weighing techniques will have some apparatus to weigh cash, by way of example a cash register drawer or a hod-type carrier, and the generic term that will be applied henceforth to all these variants will be weighing means'. It is also to be appreciated that there are a number of ways of achieving the above results, and the examples provided only illustrate one method utilising weighing techniques.

Stacking Stacking is used to remove rounding errors as banknotes are counted. When the number of banknotes is to be evaluated from the weight of a plurality of banknotes, it is more than likely that the number of notes will be found to be inexact. From this result, a simple integer rounding process is usually to employed to give an exact number of banknotes.

For example, if the weight of a bundle of banknotes to be weighed and counted indicates that 22.3 banknotes are present, it is reasonable to conclude that 22 banknotes are present and that the fractional remainder of 0.3 banknotes can be safely ignored. The value of 22 banknotes is then presented as the answer to the cashier or operator, and a cash value can be attributed to this number of banknotes and also displayed.

If another identical bundle of 22 banknotes is added to the 22 banknotes currently being evaluated, by following the same calculation it will be determined that 44.6 banknotes are present, thus giving an answer of 45 banknotes once the same integer rounding step has been effected. This answer is obviously wrong, as only 44 banknotes are present.

s Stacking recalls the fractional remainder of 0.3 banknotes - called the stacking remainder - from the first weighing and subtracts this from the second weighing to produce a 'stacked' result of 44.3 banknotes, which when rounded becomes 44 banknotes - the correct result. If subsequently more banknotes are added to the two bundles already being weighed, the stacking remainder to be carried forward to the next weighing will, in this case, be 0.6 banknotes.

Stacking is a useful procedure to employ in cases where successive pluralities of banknotes are added to ones already on the weighing means, and it significantly reduces the opportunity for errors as items are counted, particularly for larger numbers of items.

A simpler form of the above is to recall the previous weight and simply subtract the previous weight from the newly acquired weight, to provide an evaluation of incremental weight and therefore an accurate representation of the number of notes.

Try Less In the above example of stacking, should the first and second bundles of banknotes to be weighed be put on a weighing platform at the same time, the weight of 44.6 banknotes will be determined and a false reading of 45 banknotes will be recorded. This problem may be resolved by limiting the 16 number of banknotes that can be weighed at the same time to, for example, between 20 and 30 banknotes typically. Restricting the number of banknotes in this way reduces the effect of cumulative variance between the actual weight and the standard weight.

If the number of banknotes to be counted is within a maximum threshold for rounding purposes of, for example 22 banknotes, then the correct number of banknotes will be calculated. Knowing this, should the weight corresponding to more than 22 banknotes be placed on the weighing means at once, the cashier will be advised to remove banknotes until 22 or less banknotes are 26 present on the weighing means to ensure an accurate count is possible.

If a cashier attempts to count too many banknotes at any one time the cashier will be advised to remove some of the items added and try again with a lesser number of banknotes. This process will be repeated until the cashier removes enough banknotes so that the number to be weighed falls below the maximum threshold. The overall error is thus reduced and counting can recommence, for example using the stacking procedure described above.

Thus, in the example above, the Try Less method uses the total number of banknotes being weighed as the criteria for alerting the operator to 'try less' banknotes. However, the Try Less method may also use the fractional remainder of the determined number of banknotes as the criteria. To illustrate this, consider a banknote weighing action which determines that the number of banknotes present comprises a fractional remainder in a band of between, say, 0.4 to O.6 of a banknote. A fractional remainder within this band means there could almost as likely be one more banknote as one less - thus a high uncertainty factor. A fractional remainder within a band of, say 0.4 to 0.6 of a to banknote, would trigger a Try Less warning to the operator.

Techniques are also known to be able to vary the limit of the number of items that can be counted in any one operation depending on the amount of the stacking remainder. This automatically increases the number of items that can be counted if the cash item is of a sufficiently high quality- i. e. fresh banknotes which are very close to the standard weight will result in a higher maximum 'try less' threshold value.

Re-calibratina It is known to adjust, or re-calibrate, the standard weight for any type of cash item due to the varying conditions of cash items encountered during trade.

Furthermore, over some longer periods of time the actual weight may drift away from the standard weight. Counting means are known to initiate recalibration of the standard weight upon sensing a trigger action, typically a new transaction which will require a new cash count after the transaction finishes, or the action of opening and closing the cash register drawer. For to avoidance of doubt, the term "transaction" may comprise the following sample actions: a normal sale transaction, a nosale drawer open action, a shift change involving change of staff, a management report request or other similar cashier or Point of Sale terminal related activity. Known counting machines only perform recalibration actions when associated with a new transaction.

A typical re-calibration method employed incorporates first evaluating the Total Number of a type of banknote in a group, found by dividing the total weight of a group of that type of banknote by the standard weight of that type of banknote. If the resultant value is an exact whole number, there is no need to carry out the remaining steps of the re-calibration action. However, as the banknotes of the group will not usually all equate exactly to the standard weight, the resultant value will almost certainly contain a fractional part. The Mantissa of the resultant value is used in a further index calculation typically c according to the following formula: Mantissa Amp Total Number) Wherein is a calibration adjustment speed factor, and the result of the above calculation is applied to the standard weight to form the new updated and re- calibrated standard weight.

The calibration adjustment speed factor represents the interval between re- calibration actions. If the re-calibration interval is too short, erroneous operation (such as the placement of a heavy coin into a banknote compartment by mistake) will severely affect the next re- calibration before the mistake is spotted and corrected. Conversely, if the re-calibration interval is too long, too much time may have elapsed since the last re-calibration and the most recent cash value results may have been calculated erroneously. The range of appropriate values for are determined by experimental analysis, and it is usually set at the lowest possible value consistent with adequate to responsiveness, i.e. a suitably long time span between re-calibration trigger actions.

Since each counting procedure usually results in a remainder, it is possible to use this remainder to determine how far the actual weight has drifted from the standard weight, and then adjust the calibration in the direction of the actual weight. This adjustment process is performed slowly to ensure that erroneous situations, such as a wrong coin or banknote placed on the weighing means, does not affect the calibration.

go The 'try less' and stacking methods can be used effectively when banknotes are removed, although in the case of a hod-type counter it is usual for all the banknotes to be removed at once, easing evaluation difficulties.

Although the above methods are applicable to all cash counting machines c employing weighing techniques they are not without significant drawbacks.

For example, it is impractical to employ 'try less' in a cash drawer environment or a busy retail environment as it is unfeasible for cashiers to spend time after each transaction removing banknotes until the number to be weighed is below the maximum permitted threshold, although it would be possible to indicate to a cashier that this operation is required.

Also, differences in usage of cash register and hod-type counters affect effectiveness of different counting methods. For example, with cashregister drawers, banknotes are typically added in small numbers and removed in larger numbers (but leaving some in the cash drawer), coins are typically added in large quantities and then added or removed in small quantities, and vouchers or other cash items are typically totally removed from the drawer at appropriate times such as the end of a shift or the end of a working day.

Another type of problem encountered by cash register drawer type counting machines, but not by the hod-type counting machines, is that it is rare to remove all the banknotes and a small number of banknotes are usually retained in a cash register drawer to assist in the provision of change to a customer tendering a higher value banknote. The implication of this is that to current counting methods are not adequate for use in a cash register drawer scenario.

Furthermore, banknotes are often removed in large quantities (e.g. during a cash skim or when preparing cash for deposit) and there is considerable chance for miscalculation of the number of notes removed to occur, even assuming that a recent re-calibration has been undertaken. Taking a large number of banknotes off the weighing platform at once presents the same problems as adding a large number of banknotes at once, as previously described, although this problem rarely occurs in the context of the intelligent cash drawer.

It is with a view to solving these problems that a method of counting cash is presented comprising repeatedly weighing a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; after each weighing, determining the number of items in the group; comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings; as a result of the comparison, determining whether items have been added or removed; choosing a respective counting method in accordance with whether items have been added or removed; and counting in accordance with the method chosen.

In a second aspect of the invention, there is provided a method of counting cash comprising repeatedly weighing a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; and after each weighing, determining the number of items in the group; wherein the number of items in the group is determined by a plurality of counting methods each producing a respective answer.

In an alternative embodiment of the invention, there is also provided apparatus for counting cash comprising weighing means adapted to repeatedly weigh a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; processing means to adapted for determining the number of items in the group after each weighing and comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings; comparison means for determining whether items have been added or removed; choosing means for choosing a respective s counting method in accordance with whether items have been added or removed; and counting means for counting the number of items in accordance with the method chosen.

In a second aspect of the alternative embodiment, there is provided apparatus no for counting cash comprising: means for repeatedly weighing a variable- weight group of cash items as the weight of the group varies with addition or removal of cash items; means for dividing the aggregate weight of the group by the number of items in the group to obtain an updated average weight per item of the group after each weighing; storage means for storing the updated z average weight; and means for using the stored updated average weight to determine the number of items in the group following a subsequent weighing operation.

Further optional and preferred features are described in the appended So dependent claims.

In order that the invention may be more readily understood, some embodiments in accordance therewith will now be described, by way of example, with reference to the accompanying tables and figure, in which: Figure 1 exemplifies a typical hardware system for implementing the method of the invention; Figure 2 shows a schematic diagram of a first aspect of the method of the invention when implemented on the hardware system shown in Figure 1; Table 1 illustrates typical problems associated with current weighing techniques; Table 2 shows sample data demonstrating advantages of using averaged cash item weights, according a first aspect of the invention, by comparison with current weighing techniques; and Table 3 shows the method of the invention as demonstrated in Table 2, when applied to further sample data representing a more complicated scenario and also compared to current weighing techniques.

In Figure 1, there is provided weighing means 2 which may comprise the cash go drawer of an intelligent cash register or some type of weighing scale with an electronic output indicative of the weight measured. Weighing means 2 is connected in turn to a processing unit 4 so that data relating to the weight of a sample placed upon the weighing means 2 may be transmitted electronically to the processing unit 4. Also connected to the processing unit 4 is a data z store 6 and the processing unit 4 may read and write data to/from the data store 6. A video display unit 8, or some other means of alerting an operator or linked master control unit (not shown) to an erroneous situation, is optionally connected to the processing unit 4. For example, the processing unit 4 may, instead of being a stand-alone unit, be connected as part of a larger network (not shown in Figure 1) and receive and transmit data to a network controller (also not shown).

In use, to determine the value or number of a plurality of cash items placed on the weighing means 2, the weight reading output from the weighing means 2 is fed into the processing unit 4 which determines the cash value by processing the weight reading output using a stored value of the standard weight retrieved from the data store 6.

Optionally, all the weight readings determined by the weighing means 2 may be saved to the data store 6 for retrieval at the end of the day or working shift, together with timestamps relating to the recordal of each item.

It is envisaged that a cash counting device employing weighing means would to be able to constantly monitor the weight of the cash items, and detect when a transaction, or a significant change in the cash held on the weighing means, has taken place by detecting the weight difference in any one, or each, of the cash types being monitored. Typically this will then trigger a cash count. The threshold of weight change above which a count is triggered may be manually set or it may be adaptive according to the pattern of previous cash transactions.

Turning to Figure 2, the system as shown in Figure 1 has already determined from a previous count, or from a manual input, how many cash items of a particular type of cash item are present in the weighing means 2 (shown in Figure 1). The system at this point is in a wait state, waiting for an initiating - signal of a detected weight change. The initiating signal is detected at 10 as the weight of the cash item monitored by the weighing means 2 changes, and the system decides at 12 whether the change is an increase or otherwise in e weight of that cash item. Should the outcome be that the weight has increased for that cash item, representing an increased number cash items, the appropriate strategy for counting the number of cash items added is then employed at 14, for example using stacking or 'try less' methods as appropriate, the results displayed or otherwise communicated to the operator so or network controller at 16 and the stacking remainder as described previously is updated at 18 to the data store 6 in Figure 1. A new figure for the number of cash items is calculated and this integer number is used at 20 to determine a new, re- calibrated, standard weight of that type of cash item. The system then returns to a wait state whilst monitoring the weight of those cash items, waiting for another initiating signal of a weight change and for the process to restart at 10.

If, however, the determination at decision step 12 is that the number of cash items has decreased corresponding to a reduction in weight, the new number of cash items is calculated at 22 by dividing the total current weight of cash items by the most recent value of standard weight for that cash item. The results are displayed at 24 and the system then returns to a wait state whilst continuing to monitor the weight of cash items, waiting for another initiating signal of a weight change and for the process to restart at 10.

In the above description accompanying Figure 2, although mention is made of only one type of cash item, the same process is applied to all the types of cash items monitored.

From the example shown in Figure 2, it is clear that by following the invention, a cash counting device may employ more than one weighing regime and consequently benefits from a structured approach depending upon the nature and magnitude of the weight change from the previous weighing.

Further refinements include using the total stacking remainder (arrived at from the total of all the stacking remainders as the banknotes were added to the present time) and allocating a proportion of that remainder to future calculations, where the proportion is derived from the total number of as banknotes present just before any were removed. This allocation of remainder can be made on the basis of making a temporary adjustment to the calibration value or simply allocating a proportion of the remainder based on the approximate number of notes removed.

Now turning to Tables 1 to 3 - again, these only refer to banknotes for conciseness but the method illustrated is applicable to all cash items.

Furthermore, in the Tables 1 - 3 below it is assumed that the correct number of banknotes is known at the outset of the removal of banknote step. This value is known from a previous count or from manually entered data at the start of a shift or working day.

Notes Added Actual Weight Evaluated Number of Notes 1 1.02 grams 1 2 2.04 grams 2 3 3.06 grams 3 5.10grams 5 10.20 grams 10 20.40 grams 20 25.50 grams Very Uncertain 25 or 26 26 26.52 grams Wrong - 27

Table 1

Table 1 shows how current cash counting machines employing weighing techniques determine the number of banknotes present on a weighing platform using the current standard whole integer rounding technique. In this to case, and also in Tables 2 to 4, every banknote to be weighed has a standard assumed weight, before measurement, of 1 g. In this representative demonstration, the column 'Actual Weight' shows how much each number of banknotes added to the weighing platform (in the column 'Notes Added') actually weighs, and how many banknotes are considered to be present is shown in the right-most column 'Evaluated Number of Notes'. It is clear that the actual weight of the banknotes is 1.02g, representing a possible, if perhaps exceptional, +2% variance, and it is also clear that the present technique of employing simple integer rounding leads to errors after just 25 or 26 banknotes are added to the weighing platform. The calculation performed no is simply to divide the total weight in each case by the reference standard weight. In the example shown in Table 1, when 25 banknotes are added to the weighing platform the cashier performing the weighing operation would typically be notified to intervene, either by manually counting or by some other means (e.g. using the 'try less' method) as the model provides an uncertain z result if too many banknotes are counted at once.

In practice, operator intervention is typically required if any counting operation results in a number with a fractional value falling between 0.4 to 0.6 of a banknote. The result of any integer rounding which is carried out on a value falling between these two limits would be considered to be too uncertain to be reliable.

Fortunately, it is unusual to add a large number of banknotes to a cash drawer at any one time. So, the problem identified above seldom occurs as even at to the start of the shift or working day, when the cash register drawer is being prepared with the initial starting cash, it is exceptional to insert more than 25 banknotes at a time. In practice, therefore, this method is acceptable for use with intelligent cash registers which use weight sensors.

Table 2 shows a sample set of banknote data, and how the weighing method above compares to a weighing method forming part of the present invention.

One simple way to improve accuracy when dealing with the removal of a large number of notes at once from a weighing platform is to determine which number is smaller, the weight removed or the weight remaining. The smaller go weight value is then used in the weight calculation to determine either (a) how many banknotes have been removed, or (b) how many banknotes remain, from which the number of banknotes removed can then be determined in a subsequent calculation step if needs be.

s The smaller weight value of the two, i.e. of the weight removed or the weight remaining, is used because it represents a smaller number of banknotes, and hence, a smaller potential cumulative variance between the actual calculated weight of each banknote and the standard reference value. As shown in Table 2, for calculations where up to 40 notes are removed at once from the 100 banknotes originally present, the weight value used is the weight of the banknotes removed. This is simply because 40 out of 100 banknotes is less than the 60 out of 100 banknotes that remains. If, however, 50 banknotes or more are removed at once from the 100 banknotes originally present, the weight used is the remaining weight of banknotes on the weighing platform.

This is because 50+ out of 100 banknotes is equal to or greater than the number of banknotes remaining.

It is clear that minor deviations of the counting methods described previously are sufficient for the addition of items to a cash drawer, however, these same methods will be less successful during the removal of items from the cash drawer.

As banknotes are usually removed from a cash drawer in larger quantities to than they are usually added, there is a considerable chance of a miscount of the number of banknotes removed, even allowing for standard weight re- calibration. This is ameliorated by the above method of considering how many banknotes have been removed, and then calculating the number of banknotes remaining if more than half the banknotes have been removed, or calculate the number of banknotes removed if more than half the banknotes remain.

The banknotes in Table 2 are assumed to have a weight variance of +2% compared to the standard weight of 19, i.e. the banknotes in Table 2 weigh 1.029 as they did in the example shown in Table 1.

As can be seen in Table 2, calculating the number of banknotes by dividing the chosen weight value by the standard reference weight will lead to erroneous results under certain conditions. For example, removal of 40banknotes involves a weight change of 40.80 grams which, if rounded up to 41 banknotes, produces an incorrect result. This is demonstrated by the method shown in the middle column of Table 2 under the heading 'using standard reference weight'. This method simply assumes each banknote possesses an individual weight equivalent to the standard reference weight, and counting operations are conducted using this weight (which is 19 in this case) and the results shown. It is clear that this simplistic prior art method has several shortcomings as shown by the erroneous results generated.

However, using an updated average weight, determined by dividing the total weight of banknotes on the weighing platform before any banknotes are removed by the predetermined (and assumed to be correct) number of banknotes, and using this updated average weight in the counting operation yields greater accuracy in determining the number of banknotes either taken off or remaining on the weighing platform. That is to say, where 40 banknotes are removed in the knowledge that each weighs on average 1.02g, a weight change of 40.8Og will correctly indicate that 40 banknotes have been removed.

In other words, the preferred method is to use a simple average weight when to calculating the number of cash items removed. The original standard weight is ignored and a simple average weight calculated from the total number of cash items present and the total weight of the number of cash items present. The average weight will only be used when items are removed, prior to displaying and communicating the revised result. It does not matter when the averaging calculation is completed as long as it is completed between the adding of items and before the number of items removed is displayed or otherwise communicated.

It is found that using this updated average weight of banknote during bulk go removal of banknotes yields considerably more accurate results than methods based upon a standard calibration weight when the actual bank note deviates from the expected standard calibration weight.

A refinement of the preferred method would be to reconcile the newly calculated average weight against the standard weight to make sure that they approximately correspond, perhaps by including the stacking remainder. In the event of a gross error, a warning could be displayed to the cashier/operator and the data logged on a data store. The gross error margins may be determined from experience or set manually. This facility takes account of operator error, such as placing one or more coins, or other heavy items, into a drawer compartment intended for banknotes.

It is envisaged that depending upon whether the number of banknotes removed or the number of banknotes remaining is to be counted, different updated average weights may be used corresponding to the updated average weight of the banknotes remaining, or the updated average weight of the banknotes removed. In other words, different algorithms may be used depending upon the difference in cash item weight, i.e. a first algorithm if the 6 difference represents less than 50% of the initial cash item weight and a second algorithm if the difference represents 50% or more of the initial cash item weight.

As shown above, different algorithms may be optimised for different cash to weight conditions. It is furthermore envisaged that multiple algorithms may be employed during the counting procedure, perhaps in parallel, and the results from each different algorithm assessed for their accuracy. The test for accuracy may be the size of the fractional remainder discarded during an integer rounding operation, or it may be by comparison with the results from the other algorithms. For example, a majority verdict on the number of cash items may be accepted as the correct result when more than one answer results from three or more different algorithms.

Furthermore, one or more algorithms may be employed depending upon factors such as the number of banknotes added or removed, or other factor such as the end result or the fractional remainder.

Turning now to Table 3, this example also has 100 banknotes being weighed and counted, and in this example the first 50 banknotes have a +1% variance 26 and the second 50 banknotes have a +3% variance from the standard weight of 1g. As can be seen, the method using the newly calculated average weight as shown in the right-most column is able to produce correct results by compensating for the uneven variance distribution in this example. The prior art method, as shown in the middle column, is not able to compensate for all so the number of banknotes removed as shown. This represents an extreme set of circumstances that are highly unlikely to be found in practice, but is included here better to illustrate the algorithm.

A consideration to bear in mind is that even in a case wherein one half of banknotes to be weighed has one standard weight, and the other half has a slightly different standard weight, it can be shown that with 100 banknotes to be weighed a percentage deviation of 2% is allowable, and if 200 banknotes 6 are to be weighed (an unusually large number), that deviation falls to an allowable 1%. Both these deviation amounts are unlikely in reality.

It is also important to consider the effect on accuracy of mistakenly placed items, for example, a coin being placed in a banknote weighing compartment.

to It must be accepted that the addition of a coin will cause the weighing system to miscount, and this is unavoidable. However, the maximum amount that the coin will affect the average weight of a banknote (assuming that a miscount has occurred) is only half a banknote. This results from the rounding process as any miscount arising from, say, the erroneous placement of an item will 16 result in a miscounted integer value and/or miscounted mantissa value. The maximum value of the mantissa miscount will be 0.5 of a banknote as any greater value will affect the integer miscount. Thus it is seen that if more than one banknote exists on the weighing means the overall average weight of banknote is unaffected In this way, the effect of spurious items is significantly go limited in terms of affecting future averaging or calibration.

Claims

Claims 1. A method of counting cash comprising: repeatedly weighing a

variable- weight group of cash items as the weight of the group varies with addition or removal of cash items; after each weighing, determining the number of items in the group; to comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings; as a result of the comparison, determining whether items have been added or removed; choosing a respective counting method in accordance with whether items have been added or removed; and counting in accordance with the method chosen.
2. The method of Claim 1, comprising: choosing a respective counting method in accordance with the number of items that have been added or removed between successive weighings; and counting in accordance with the method chosen.
3. The method of Claim 2, wherein the counting method chosen depends upon the ratio of items added or removed to the items previously in the group.
4. The method of any preceding Claim, comprising: after each weighing, dividing the aggregate weight of the group by the number of items in the group to obtain an updated average weight per item of the group; storing the updated average weight; and using the stored updated average weight to determine the number of items in the group following a subsequent weighing operation.

Jo
5. The method of any preceding Claim, wherein a fractional remainder produced upon determining the number of items following the first of successive weighings is stored and applied in whole or part to determining the number of items following the second of successive weighings.

1s
6. The method of Claim 5, wherein a proportion of the fractional remainder is applied to determining the number of items following the second of successive weighings, which proportion is determined from the total number of items determined following the first of successive weighings.
7. The method of Claim 6, wherein the proportion of the fractional remainder is based upon the ratio of items added or removed to the items previously in the group.
8. The method of any of Claims 5 to 7 when appendant to Claim 4, wherein as the fractional remainder is applied as a temporary adjustment to the stored average weight.
9. A method of counting cash comprising: repeatedly weighing a variableweight group of cash items as the weight of the group varies with addition or removal of cash items; and after each weighing, determining the number of items in the group; wherein the number of items in the group is determined by a plurality of counting methods each producing a respective answer.
l O. The method of Claim 9, wherein the answers produced by the respective counting methods are assessed and an appropriate answer is adopted.
11. The method of Claim 10, wherein the answers of three or more counting lo methods are polled and an answer is adopted if it corresponds to a majority of those answers.
12. A method of counting cash comprising: 1 repeatedly weighing a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; after each weighing, dividing the aggregate weight of the group by the number of items in the group to obtain an updated average weight per item of the group; storing the updated average weight; and using the stored updated average weight to determine the number of 2s items in the group following a subsequent weighing operation.
13. The method of Claim 12, comprising comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings.
14. The method of Claim 12 or Claim 13, comprising comparing the updated average weight with a previously calculated average weight and discarding the updated average weight if the difference between them exceeds a predetermined threshold.
15. The method of Claim 14, wherein the threshold is determined in accordance with a fractional remainder produced upon determining the number of items following a weighing.
16. Apparatus for counting cash comprising weighing means adapted to repeatedly weigh a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; processing means adapted for determining the number of items in the to group after each weighing and comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings; comparison means for determining whether items have been added or removed; choosing means for choosing a respective counting method in accordance with whether items have been added or removed; and counting means for counting the number of items in accordance with the method chosen.
17. The apparatus of Claim 16, further adapted to include choosing means for choosing a respective counting method in accordance with the number of items that have been added or removed between successive weighings; and counting means for counting in accordance with the method chosen.
18. The apparatus of Claim 17, wherein the counting means is further adapted to count according to the method chosen respective to the ratio of items added or removed to the items previously in the group.
19. The apparatus of any preceding Claim, further adapted to provide processing means able to divide the aggregate weight of the group by the number of items in the group to obtain an updated average weight per item of the group after each weighing; storage means for storing the updated average weight; and processing means to determine the number of items in the group following a subsequent weighing operation using the stored updated average weight.
20. The apparatus of any preceding Claim, further adapted to comprise means wherein a fractional remainder produced upon determining the number of items following the first of successive weighings is stored and applied in whole or part to determining the number of items following the second of successive weighings.
21. The apparatus of Claim 20, further adapted to comprise means wherein a proportion of the fractional remainder is applied to determining the number of items following the second of successive weighings, which proportion is determined from the total number of items determined following the first of successive weighings.
22. The apparatus of Claim 21, further adapted to comprise means wherein the proportion of the fractional remainder is based upon the ratio of items added or removed to the items previously in the group.
23. The apparatus of any of Claims 20 to 22 when appendant to Claim 19, further adapted to comprise means wherein the fractional remainder is applied as a temporary adjustment to the stored average weight.
24. Apparatus for counting cash comprising means for repeatedly weighing a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; and means for determining the number of items in the group after each weighing; means for determining the number of items in the group by a plurality of 6 counting methods each producing a respective answer.
25. The apparatus of Claim 24, further adapted to comprise means for adopting an appropriate answer from assessing the answers produced by the respective counting methods.
26. The apparatus of Claim 25, further adapted to comprise means for polling and adopting an answer if it corresponds to a majority of the answers from three or more counting methods.
27. Apparatus for counting cash comprising: means for repeatedly weighing a variable-weight group of cash items as the weight of the group varies with addition or removal of cash items; means for dividing the aggregate weight of the group by the number of items in the group to obtain an updated average weight per item of the group after each weighing; 2 storage means for storing the updated average weight; and means for using the stored updated average weight to determine the number of items in the group following a subsequent weighing operation.
28. The apparatus of Claim 27, further adapted to comprise means for comparing the number of items determined following the second of successive weighings and the number of items determined following the first of successive weighings. 25
29. The apparatus of Claim 27 or Claim 28, further adapted to comprise means for comparing the updated average weight with a previously calculated average weight and discarding the updated average weight if the difference between them exceeds a predetermined threshold.
30. The apparatus of Claim 29, further adapted to comprise means for determining the threshold in accordance with a fractional remainder produced upon determining the number of items following a weighing.