Evaluation of the Hendon Arm: Report of a Medical Research Council ad hoc Working Party


The "Hendon" arm was designed between 1963 and 1969 by the Medical Research Council Powered Limbs Unit at West Hendon Hospital, under the direction of Dr. A. B. Kinnier Wilson. In 1969 the Council set up a multidisciplinary Working Party to plan, coordinate, and supervise a full evaluation of this prosthesis and to report to the Council on the results. This paper is an abridged version of that report.

Design and Construction of the Hendon Arm

The Hendon arm was intended for use by children with severe upper-limb deficiencies. It was designed to be as natural in size and appearance as possible (with the exception of the terminal device which was of the split-hook type) and to lift a mass of 1 kg. Plastics were used in the construction of the arm to save weight.

The bilateral prostheses and harness ( Figure 1-A and Figure 1-B ) weighed 3.9 kg excluding the power supply. There were five modules, each of which included an actuator.

The controls were specifically designed for each individual child and combined in a device resembling a joystick ( Figure 2-A and Figure 2-B ). A system of position feedback transmitted to the control lever information on the movements of the prosthesis. The arms were powered by carbon dioxide at a pressure of 250 lb/in2, and stored in cylinders, each containing 180 gin. Each cylinder lasted for one hour of moderate activity with bilateral prostheses 2,3 .

Clinical Evaluation Plan


Very few clinical evaluations of powered artificial limbs had been conducted and the Working Party was faced with the task of not only evaluating a prosthesis, but also of designing a method by which this evaluation could be carried out. It was decided to assess: 1) the extent and quality of a child's performance with the prosthesis and whether this performance improved with practice; 2) the time and effort needed to learn to control the device; 3) the amount of control that could be achieved; 4) whether the benefits provided outweighed the disadvantage of wearing the arm; and 5) whether the child sometimes managed better without the artificial limbs, perhaps by foot use.

Centres Selected for the Trial

It was considered essential that the centres participating in the evaluation should have clinical charge of children needing powered prostheses; and that there be on-the-spot technical facilities for the maintenance of the arm. At the same time the centres had to have ready access to the Hendon Unit. Two centres which met these criteria were those at Chailey and at Oxford.

The Selection of Children

Children who needed a powered prosthesis, and who had upper-limb rudiments appropriate to the Hendon type of joystick, were considered for the trial. One child with upper-limb amelia was accepted as he could move the joystick with his feet, despite a lower-limb deficiency.

It was vital that the children be available at the appropriate times for fitting, training, and review. Moreover the parents of the children living at home had to be willing to take part on this basis. Social problems excluded some children who would have been suitable from the physical point of view. Only six children who fulfilled all the physical and social criteria were found.

Each child's limb deficiencies and other physical disabilities, such as spinal abnormalities 5 , dwarfism7, or heart disease8 were recorded. These data are available from the authors on request.

Nichols, Rogers, Clark and Stamp6 have emphasized the need to study the psychological considerations involved in the rejection of powered limbs. The large number of disciplines involved in the management of the limb-deficient child and the effect of this multiplicity on rapport have been noted 4 . Hutt' has pointed out that the factor which limits performance might be lack of understanding of the task required rather than inability to control a prosthesis.

The intellectual and perceptual abilities of the children were assessed. When necessary, the homes of the children were visited and family and environmental backgrounds were recorded.

The Fitting and Training Program

Previous experience indicated that a minimum of six weeks would be required to make and fit a control system for each subject.

A period of six weeks was also assigned for the training of each child in the use of the prostheses. Training exercises were progressive in nature, incorporating tasks which would be useful to the child.

Functional assessment was carried out using both structured and unstructured tests. The 31 structured tests were selected to assess the ability of the child to control particular modules of the prostheses. Five of the tests required the use of both arms simultaneously. The 45 unstructured functional assessment tests were organized in terms of activities of daily living, and allowed a trained observer to assess the child's attitude toward the prostheses and to evaluate his performance.

Performance was graded 3, 2, 1 or 0 according to whether the test was performed "smoothly," "hesitantly," "with considerable difficulty," or "not at all."

Results of the Clinical Evaluation

None of the six children was able to complete the evaluation as planned. There were two reasons for this: the modifications and repairs necessary to the prostheses, and the eventual unacceptability of the test regime to the children.

Fitting the Arms

Table 1 lists the children in order of their degree of physical disability, and gives their age at the time of fitting. It also gives the length of the fitting period, the number of visits and amount of time involved, and the number of modifications or repairs needed.

These data show that fitting usually took more than the anticipated six weeks. The extended fitting periods were mainly due to the frequent need for modifications and repairs. Delays caused by school holidays and the need to arrange fitting sessions in non-school hours were contributing factors.

Use Training

The training period for each child, the actual number of sessions involved and their average duration, and the number of modifications or repairs to the arms during the evaluation period are also given in Table 1.

Two children completed training very quickly in two or three weeks, whereas two others took about five months.

It had been hoped that most of the technical difficulties would be overcome during the fitting period, and that the training period would be relatively troublefree but, as Table 1 indicates, technical problems continued to develop after the arms had been fitted. In every case there were some interruptions in the training program because of mechanical breakdowns. These tended to occur during the early stages of training.


Assessments of Progress During the Training Period

The structured assessments were made at the beginning, middle, and end of the training period, and the unstructured assessment at the end of this period.

The results given in Table 2 show a gradual increase in the average score for the structured assessments, indicating a clear improvement in the facility with which children handled the arms. On sequent trials, the children succeeded in doing more of the tests and in doing them better. The improvement in performance scores is shown graphically in Figure 3 .

Status at the Completion of Training

By the end of the training period most of the children had become tired of performing the set tests, and refused to complete either the structured or the unstructured assessments at the one- and three-month reviews.

In summary, two children rejected the arms outright but for different reasons. Subject two was very intelligent and the arms did not do enough for him, whereas subject six had a relatively low intelligence level, had little powers of concentration, and was easily distracted. Subject five used the arms as a "tool" for carrying things about. One child (subject four) had a positive attachment for the devices, even though they appeared not to provide any functional advantage. Two children (subjects one and three) obtained some useful function from the arms. However, when reviewed again 15 months after the end of the training period, subject one had stopped using the prostheses.

All the children found that the arms interfered with the performance of certain activities which they could do better without the prostheses. Although they gave added reach, the prostheses were slower than the children's own limbs. It was found that the arms would not enable children to manage their personal toilet, in particular the problems of defecation and menstruation which require a combination of reach and manipulative skills.

Unilateral and Bilateral Activities

The average performance scores for the unilateral and bilateral activities are compared in Table 3 and Figure 4 . Most children were reasonably dexterous with one powered arm, but immediately they tried bilateral activities they got into difficulties. Many unstructured activities which seem to be bilateral are in fact unilateral activities performed with one arm and then the other in sequence.

After each session with the children the therapists recorded any relevant comments which they or the children had made, and a considerable amount of material was accumulated. These subjective comments were as valuable as the results of the formal structured and unstructured assessments and, although they were not amenable to statistical analysis, they provided the only record of progress for those children who refused to complete the formal tests.


Balance was a problem for those children who operated their powered arms while using lower-limb prostheses, and they preferred to use the pneumatic arms when they were seated. The child with lower-limb amelia used the arms while standing in a bucket-type lower-limb device which overcame the balance problem. The child with upper-limb amelia who used the arms by means of a foot-operated "console" also had balance problems.


Parents, teachers, and therapists frequently criticized the weight of the prostheses, but the children themselves did not complain very much about it. Comments were made occasionally that the arms were "heavy," but later in the study the children said that the arms didn't feel as heavy when they got used to them. On the other hand, the arms were not comfortable to wear.


In the early stages of the evaluation, parents and teachers criticized the appearance of the arms, but soon became used to them. The children themselves made no such comments.

Results of the Engineering Evaluation

Laboratory Aspects

The designers of the Hendon arm sought to achieve a high level of sophistication with respect to the number and range of controlled movements, the type of control, the dynamic response and stall loads, and the design of the linkages. High stall loads and good dynamic response were achieved by the use of a higher gas pressure than heretofore, and this involved the development of special connectors and a careful choice of gas tubing. The linkages and actuators were designed on a modular concept; this not only facilitated maintenance but also permitted one or more modules to be employed if the complete arm prosthesis were not required. The design specifications, coupled with the usual weight and size restrictions, presented a formidable problem and, in attempting a solution, unusual materials, techniques, and designs were used.

To reduce the weight of each module to the bare minimum, modern plastics with high strength/weight ratios were used. However, the particular material chosen proved to be both brittle and subject to abrasive wear. Thus in test conditions some components failed under shock loading where the arm was mounted rigidly. This might not have happened if the arms had been mounted on a harness which in turn sat on a shock-absorbing human body, and indeed no such failures occurred in clinical use. Where the material was used as a bearing surface, such as in the elbow linkage and in the wrist rotator, wear soon caused play in the mechanism. The other serious drawback encountered was that of water leakage. It was necessary to use water damping to achieve stable control. Unlike a small gas leak, which though wasteful of power did not otherwise affect the system, water leakage rapidly restricted the range of motion of the actuators; also, gas leakage into the water system could cause instability. Thus frequent attention to the water reservoir was required if any leakage into or out of the water system occurred. There is some doubt as to whether a good seal can ever be achieved in a system with high-pressure gas on one side of a piston and hydraulic fluid on the other. A solution to the problem of preventing gas from seeping into the water lines would be to use a double seal in the piston with a chamber vented to atmosphere in between. This is accepted practice in such a situation but its incorporation into the Hendon actuators would involve a major redesign. With the current single-seal system, the compressed gas tended to remove any lubricant which was put into the piston, and the wear on the seals was accelerated.

In the elbow and humeral rotator, the use of a nylon piston head resulted in a buildup of friction forces due to swelling of the nylon by water absorption.

In many instances, malfunctioning of the arm occurred because of inadequate location and locking of various parts and components.

Minor redesign would remove many of these faults.

The speed of response and stall loads of the shoulder-flexion, elbow-flexion, and wrist-rotation units were measured. During clinical trials the motions of the shoulder flexion, humeral rotation, and elbow flexion were slowed down by the use of restrictors. If it becomes apparent that the highest response speeds will not be used, it may be better to use lower flow-rate valves which give a potentially more controllable system.


This was the first known occasion on which pairs of five-movement arms, with four of the movements in each arm having closed-loop control, had been fitted to a series of children with bilateral upper-limb deficiencies. It proved to be extremely difficult to harness five independent controls so that any two or more movements could be controlled simultaneously. Although the residual hands were used if possible, it became necessary to use other parts of the body for control in certain cases. A great deal of technician time was required to achieve satisfactory fittings.

The difficulty in harnessing suitable movements was further aggravated by the variability of the force levels required between sets of arms and within one set at different times.

Reliability and Maintenance

The majority of the faults which occurred during the trial were of a comparatively minor nature and most of the individual repairs required less than two hours of technician time.

Nevertheless it must be recognized that each repair, however trivial, represented an interruption to the patient's activity and caused a loss of confidence in the equipment. In this respect, frequent minor faults, even though they required only one or two hours of workshop time, were more disruptive than an occasional major breakdown.

The main areas of trouble were (1) seals, replacement being necessary both because of wear, and the difficulty in maintaining an effective seal between the water and high-pressure gas; (2) control cables, which required frequent adjustments due to the variable forces; (3) valve discs, which were sensitive to small particles of dirt and also tended to debond between the rubber and metal; (4) the terminal device, which had a wide range of faults.

Major Faults Detected

Of the two problems identified in the bench tests, only one-that of leakage between the gas and water-proved to be troublesome during the period of the clinical evaluation, and even this was contained within the technical resources provided. Wear (the other problem) certainly took place, both in the elbow-linkage and wrist-rotation units, but there was no evidence to show that this wear adversely affected performance.

General Discussion

The Working Party was faced with the fact that no adequate or objective evaluation of an artificial arm or similar complexity had been carried out previously. The present trial is one of the first in which the equipment was taken out of the hands of the design team and put into a clinical environment.

In the two clinical centres selected, only six children who met the fairly exacting criteria for participation in the trials were found. Five of these had comparable intelligence levels in the subnormal range; all six needed much support and encouragement from the occupational therapist. The small number of children limited the statistical validity of the results. Table 1 shows how each child was affected by the processes of fitting, training, and testing. The time taken reflected not only the child's competence and interest but also his availability for the trial program which was often interrupted because of schooling, holidays, illness, and even operations. Other adverse factors were delays due to staff changes and the fact that the arm needed frequent repairs. The technical repair work done on each arm was considerable.

Children with severe limb deficiencies are relatively uncommon, and for this reason most children in this category have used other power-assisted devices previously. This experience produces a degree of sophistication in both child and staff which may affect their reaction to a new device in a way which is very difficult to quantify.

All the children learned to use their limbs within two to twenty weeks and during this period were prepared to use them for all tests. To follow the children's subsequent progress, two types of tests were used, structured and unstructured. Failure to perform one or more of the structured tests often led to loss of confidence and a tendency to reject the arm completely. It was found that when the children were allowed to choose their own activities they were much more prepared to collaborate and practice use of the arms. In any event only small differences were found in the scores of the structured and unstructured tests. This suggests that in any future trials it might well be unnecessary or even a disadvantage to set up an elaborate scheme of structured tests which can be a source of considerable frustration and boredom to the child.

The original concept of the Hendon arm was for a prosthesis with large-force and fast-response capability. However, neither capability was fully realized as the arm was operated either at reduced gas pressure or with flow-limiting restrictors in the supply lines. These restrictors were fitted for the initial training period to reduce the speed of movement and were not subsequently removed as it was considered that the reduced power was adequate. Thus, parts of the original design specification have been shown to be more ambitious than was required for the particular group of patients involved in the trial; whether this would be true for an adult group of patients is a matter for debate. Had the Hendon team had greater access to patients during the initial development stages, it is possible that the specification might have been revised and for this reason the Working Party was of the opinion that future projects of this type should generally be carried out in establishments where day-to-day clinical contact can exist between the design team and patients.

The arm was clearly designed with a view to keeping its weight within acceptable limits. No doubt further weight-savings are possible, though it is difficult to identify where major savings could be made except perhaps with the water-damping system which could be eliminated if an alternative method of stabilization were found.

Some delay in the supply of arms was caused by the need for modifying or hand-fitting a number of components. This situation is typical of a prototype product, and such problems would normally be eliminated in the development stage. The Hendon team were required to produce a small batch of arms without going through such a stage. For this reason the arm should not be directly compared with a fully developed product.

The evaluation program was frequently delayed because of the development of minor faults which had to be corrected, and sometimes even necessitated modification of the design of the prosthesis. This contributed to a loss of confidence in the arms on the part of most of the children.

The results of the engineering evaluation correlated well with the various causes of breakdown experienced in clinical use, and the somewhat adverse results of this trial would have been better if the bench-testing, with any necessary design modifications, had been carried out before the arms were evaluated on the children.

In spite of these reservations, much credit should go to the design team for the level of performance that has been achieved with this prosthesis. The task faced was a severe one. The restrictions on weight, space, and energy consumption, in conjunction with the requirement for precise control of five independent movements with good reliability, created a great problem. When one couples these considerations with the need to produce a device that can be worn by a patient and replace part of the function of a missing limb with acceptable static and dynamic cosmesis, it is not unreasonable to note that this design problem has not been solved anywhere.


  1. A procedure for evaluation of powered upper-limb prostheses was devised by a multidisciplinary team.
  2. This procedure was used to assess the Hendon arm.
  3. Of six children who were fitted with the arms and trained to operate them, two gained lasting useful function from the prostheses.


Thanks are extended to all who collaborated in this evaluation, including the children, their families, and schools. The clinical psychologist, the technicians and occupational therapists, all made very marked contributions.

In particular we wish to thank Dr. Kinnier Wilson and his staff at the M. R. C. Powered Limbs Unit at West Hendon Hospital for their willing cooperation at all stages of the study.

1. Hutt, S., Some problems in the assessment of children with upper-limb prostheses. Personal Communication, 1966.
2. Kinnier Wilson, A. B., S. R. Montgomery, and R. McWilliam, Group on Power and Control Systems for Upper Limb Prostheses. Progress Report No. 1, Medical Research Council, 1966.
3. Kinnier Wilson, A. B., S. R. Montgomery, R. McWilliam, and B. D. Sanderson, The Design and Development of an Experimental Externally Powered Upper Limb Prosthetic System. Progress Report No. 2, Medical Research Council, 1970.
4. Nichols, P. J. R., MRC Conference on Powered Prostheses, 5th December 1966, MRC 66/ 1073.
5. Nichols, P. J. R., J. L. Boldero, I. W. Goodfellow, and A. Hamilton, Abnormalities of the vertebral column associated with thalidomide induced limb deformities. Orthopaedics: Oxford, 1,71, 1968.
6. Nichols, P. J. R., E. E. Rogers, M. S. Clark, and W. G. Stamp. The acceptance and rejection of prostheses by children with multiple congenital limb deformities. Artif. Limbs, 12:1, Spring 1968.
7. Quibell, L. P., Dwarfism in thalidomide damaged children. BRADU Bulletin, 3, 45, 1970.
8. Smithells, R. W., Defects and disabilities of thalidomide children. Brit. Med. J., 1, 269, 1973.