A Design Theory for Pervasive Information Systems

Panos E. Kourouthanassis and George M. Giaglis

Department of Management Science and Technology

Athens University of Economics & Business

47A Evelpidon & 33 Lefkados St., 11362, Athens, Greece

Abstract. Pervasive Information Systems (PIS) constitute an emerging class of

Information Systems where Information Technology is gradually embedded in

the physical environment, capable of accommodating user needs and wants

when desired. PIS differ from Desktop Information Systems (DIS) in that they

encompass a complex, dynamic environment composed of multiple artefacts

instead of Personal Computers only, capable of perceiving contextual

information instead of simple user input, and supporting mobility instead of

stationary services. This paper aims at proposing a design theory for PIS. In

particular, we have employed [46]’s framework of Information Systems Design

Theories (ISDT) to develop a set of prescriptions that guide the design of PIS

instances. The design theory addresses both the design product and the design

process by specifying four meta-requirements, nine meta-design considerations,

and five design method considerations. The paper emphasises mainly on the

design theory itself and does not address issues concerning its validation.

However, in the concluding remarks we briefly discuss the activities we

undertook to validate our theoretical suggestions.

1 Introduction

Information technology (IT) artefacts are already embedded in more places than just

our desktop computers, providing innovative services in ways unimaginable in the

near past. This shift in the viewpoint of information systems (IS) is commonly

referred to as ‘post-desktop’ [21] or ‘ubiquitous’ computing [48]. This trend has fired

a shift away from computers towards computerised artefacts. A new generation of

information appliances has emerged [38], differing from traditional general-purpose

computers in what they do and in the much smaller learning overhead they impose on

the user. Instead of having IT in the foreground, triggered, manipulated, and used by

humans, nowadays we witness that IT (irrespectively whether it comprises of

computers, small sensors, or other communication means) gradually resides in the

background, monitoring the activities of humans, processing and communicating this

information to other sources and intervening should it be required. This new class of

IS is often called ‘Pervasive Information Systems’ (PIS) [9] and enables new

interaction means beyond the traditional desktop paradigm.

Specifically, DIS through their form of personal computers or other stationary

access devices, were designed to fit into an office environment and the activities

E. Kourouthanassis P. and M. Giaglis G. (2006).

A Design Theory for Pervasive Information Systems.

In Proceedings of the 3rd International Workshop on Ubiquitous Computing, pages 62-70

DOI: 10.5220/0002503700620070

 SciTePress

taking place there. They were designed to be efficient tools in the hands of

professionals. Thus, their practice of interaction design is directed towards this

setting. Moreover, these systems are designed for use; this means that their design and

evaluation are accomplished on the basis of some definition of their functionality and

perceived usage. Thus, designers seek a solution that satisfies the basic criteria for

usability such as efficiency in use, low error rate, and support for recovery from error,

based on a general knowledge about what to do and what not to do to meet such

criteria [16, 32]. The objective is to achieve maximum usability with respect to a

general, precise notion of use, and the design is motivated by this ambition.

Consequently, the existing design approaches for DIS follow the same rationale.

Design methods such as SSADM [47], ETHICS [31], SSM [10], or Object-Oriented

Analysis [29] to name but a few popular methods, consider systems that support

predefined tasks and in many cases assume a job or office environment. Moreover,

such methods rely on the knowledge of the designer in order to recognise potential

problems and mostly offer a generic approach to design. Also, such methods tend to

focus on details of systems, something which is possible with traditional, static

systems, but which cannot always happen for the dynamic and rich environments

supported by pervasive systems. Finally, all the traditional approaches are oriented

towards the fundamental tenets of Human-Computer Interaction: design for a specific

user, performing a specific task, in a specific domain [36].

This position paper aims at proposing a methodological approach that may

facilitate designers to develop PIS instances. Specifically, we will present a design

theory for PIS. The following section presents the design challenge of PIS. Section 3

briefly introduces the methodological framework that will be used to specify the

design theory, while section 4 presents the theory itself. The final section concludes

with a critical appraisal of the proposed design theory and its practical usefulness.

2 Information Systems Design Theories

To specify our design theory we have followed the framework of Information

Systems Design Theories (ISDT) that was first articulated by [46]. An ISDT aims at

the design of classes of Information Systems, rather than the development of specific

IS instances. To this end, design theories are prescriptive, in the sense that they

provide constructs and guidelines for the achievement of stated goals, rather than

explaining phenomena (explanatory theories) or predicting outcomes (predictive

theories). Finally, design theories are theories of procedural rationality [42], as their

objective is to prescribe both the properties that an artefact should have if it is to

achieve certain goals, and the methods of artefact construction.

According to [46], design theories have two aspects; one dealing with the product

of design (the artefact that will form the outcome of applying the design theory) and

another dealing with the process of design (i.e. the method by which the design

product can be realised). We will use this distinction to describe the components that

form an ISDT. These components are summarised in the following table.

Table 1. ISDT Components.

DESIGN PRODUCT

Kernel Theories Theories from reference disciplines that govern design

requirements

Meta-Requirements The class of goals to which the theory applies

Meta-Design A class of artefacts hypothesised to meet the meta-

requirements

Design Product Hypotheses Used to test whether the meta-design satisfies the meta-

requirements

DESIGN PROCESS

Kernel Theories Theories from reference disciplines that govern the design

process

Design Method Description of procedures for artefact construction

Design Process Hypotheses Used to test whether the design method results in an artefact

consistent with the meta-design

The framework has already been used to develop design theories for, amongst

others, Executive Information Systems (EIS) [46], Decision Support Systems [22],

Group Decision Support Systems [27], Organisational Memory Information Systems

[43], Simulation Systems for IS Evaluation [15], and Emergent Knowledge Processes

[28]. The following section presents the proposed design theory for PIS.

3 A Design Theory for Pervasive Information Systems

3.1 Meta-Requirements Elicitation

The first meta-requirement refers to the profile of prospective PIS users. In the PIS

context it is highly unlike for the system designer to know in advance the types of

people who will be using the system. Take into consideration the users of the different

implementations of tour guides existing in the PIS literature [1, 5, 6, 11]; users may

range from people that are vaguely familiar with IT (mainly due to their interaction

with commonplace IT artefacts such as mobile phones) or (at extreme cases) techno-

phobic. Moreover, these types of users are opportunistic in the sense that they will use

the tour guides for a particular time frame and for a particular reason (in this example,

to augment their visiting experience). To this end, it is highly unlikely that these users

will be subject to thorough training in the system’s use as in the case of DIS users.

Conclusively, a PIS instance should support all the different user types by employing

sufficient mechanisms that enhance or facilitate user interactions with the system,

while at the same time haste users’ learning curve for using the PIS.

The second meta-requirement refers to the PIS capability to support the multitude

of different device types that may participate in the pervasive environment. [18]

distinguish among four types of devices: information access devices, intelligent

appliances, smart controls, and entertainment systems. Nevertheless, ideally a PIS

should support any device that has built-in active and passive intelligence. Therefore,

devices’ heterogeneity is the most important element that should be addressed during

the design of a PIS. Moreover, the plurality and diversity of pervasive devices

generate additional system requirements in terms of connectivity and integration

between them [40]. Furthermore, the design of the application and user interface

should take into account the unpredictability of end devices. In the case that an

application follows the user and moves seamlessly between devices, it is implied that

this application will have to adapt to changing hardware capabilities (different types

of pointing devices, keyboards, network types, and so on) and variability in the

available software services [3].

Designing for manipulation of contextual information, implies that a PIS should be

able to perceive relevant information of its environment (with location sensitivity and

user identity capturing being the minimum requirement as stated by [2, 12], process it,

and adapt to changes in the environment taking into account both historical and

current data. Although at present contextual information refers mainly to the users’

current location, we expect that in the near future PIS will be able to perceive

simultaneously multiple stimulants that may be contradictory one to the other. Thus,

this meta-requirement suggests that PIS should accommodate an appropriate

mechanism that will filter the different contextual information particles, process them,

and adjust their behaviour according to the information that best suits the current

occasion.

The final meta-requirement suggests that pervasive artefacts should be ‘gracefully’

embedded in the physical space. This smooth integration does not suggest that these

IT artefacts should be completely invisible to the system users, as implied by most

visionary research papers in the field [33, 41, 49, 50]. On the contrary, we follow

[37]’s considerations that pervasive technology should be governed by meaningful

presence, promoting unobtrusiveness. Thus, the challenge is to design PIS in such a

way that users perceive them as part of the environment. Universal design principles

[44] may be applied to create remembrances allowing for system usage with the

minimal distraction. Likewise, the PIS designer should also focus on the aesthetical

qualities of the pervasive artefacts [13].

3.2 Meta-Design Elicitation

The first meta-requirement clearly suggests that PIS should support opportunistic or

inexperienced users that do not have the luxury, or time, for training to the system’s

functionality. To this end, PIS designers should devise means that facilitate the

interaction of such users with the system, and minimise its learning curve. The

solution to this problem is to employ natural, easy to use and easy to learn interfaces

that facilitate a richer variety of communications capabilities between humans and

computation artefacts. At the same time, PIS designers should expect that the cases of

PIS misuse will be increased compared to DIS. This is the result of the potential PIS

users’ profile (lack of experience and/or sufficient training). Consequently, the system

design should incorporate appropriate mechanisms that minimise the degree of errors,

or guide the user in such a way that prevents errors from even occurring. Finally, PIS

should be able to perceive users’ current skill level through both the contextual

information and current system usage and adapt their functionality accordingly.

The diversity and plurality of pervasive devices poses new challenges for

information delivery applications in this environment. To meet the demands in this

heterogeneous environment, it is necessary for the information to be customized or

tailored according to the user's preferences, client capabilities and network

characteristics [19]. As such, PIS should incorporate a sufficient adaptation system

that can accommodate all different types of adaptations between different formats.

Already several such mechanisms have been proposed in the literature [8, 14, 17, 26,

30, 35] providing the system designer with the option to select the most appropriate

for its system requirements. Moreover, a PIS should be scalable. Scalability refers to

the ability to incrementally increase the abilities of a system, whilst maintaining, or

improving, performance [40]. As such, the issue of scalability refers mainly to

ensuring smooth and unobtrusive communication among PIS clients and backend

hardware infrastructure (such as sensors and actuators, backend systems, and so on).

Contextual management implies that sensing artefacts should be able to effectively

communicate the information they collect and process as well as trigger events that

deem necessary to support PIS users. Opposed to DIS where the user initiates the

interaction with the system, the PIS vision for invisibility and unobtrusiveness

suggests that the system is always active, continuously collecting contextual

information, and pro-acting (rather than re-acting) to the needs and demands of the

end users before they even start expressing them. Conclusively, the PIS designer

should devise an appropriate mechanism that supports proactive system operation.

Similarly, since it is extremely difficult to program each participating device and

application in the pervasive environment to receive and communicate uniformly the

information it collects, PIS designers should devise a representation format that is

efficient enough to model, process and communicate context.

The final two meta-design considerations stem from the requirement of smoothly

embedding the pervasive artefacts to the physical space. On the one hand, PIS should

be easily accessible. If designers follow the extreme suggestion to completely hide the

IT infrastructure, we might end up with a system that is completely inaccessible due

to the fact that users are unaware of how to use it or, at extreme cases, of its existence.

On the other hand, pervasive artefacts should be smoothly embedded in the physical

environment. As such, PIS designers should make sure that the systems which they

create do not conflict with or challenge the architecture of the place they will be

integrated. To reach that goal, co-operation between two previously completely

different disciplines (namely IS software engineering and civil architecture) seems to

be the logical solution. Consequently, the design of PIS should exploit the existing

material of the physical environment and gracefully embed ITin the physical world.

3.3 Design-Method Elicitation

The first design method consideration suggests that PIS designers orchestrate the

design around the informal and unstructured activities that users perform. This is also

illustrated in the functionality of various PIS implementations. For example, domestic

PIS, in their multiple instantiations, support such activities as home automation

(inventory management, light and heat adjustment, and so on), or home entertainment.

The focus on activities, as opposed to tasks, is a crucial departure from traditional

HCI design. Of course, activities and tasks are not unrelated to each other. Often an

activity will comprise several tasks, but the activity itself is more than these

component parts. The challenge in designing for activities is encompassing these

tasks in an environment that supports continuous interaction.

The second design method consideration suggests that prototyping through its

various forms (sketching, low-fidelity designing, and mock-ups, just to name as a

few) should be a core activity of the design process. Demonstrating prototypes to the

system users will ensure the implementation of user-friendly interfaces, as well as the

incorporation of user feedback regarding several dimensions of the system such as

usability, functionality, privacy protection, and so on. Moreover, since PIS have

significant impact to the physical environment (through the embedment of several

pervasive artefacts), prototyping may assist designers to fix functional user

requirements without implementing the PIS on a full-scale basis, thus, minimizing

implementation costs in terms of time and financial resources.

The third design method consideration suggests integrating conceptual design in

the early phases of the design process. Contextual design (CD) may be used in the

form of sketching or conceptual scenarios to demonstrate the system’s functionality to

the system’s stakeholders. Given that PIS may re-engineer the way users perform

their tasks and activities, CD may be employed as a technique that presents alternative

usage scenarios, technological solutions, or interaction techniques, so that system

users may evaluate them, and select the most user-friendly or appropriate, meeting

their goals and aspirations.

The fourth design method consideration aims at addressing issues relating to user

privacy. Indeed, context-awareness implies that the system will be able to monitor

and process personal information such as the users’ current location, activities, even

information related to the human body (e.g. users’ temperature, heartbeats, or

respiration levels). Combining these features with the capability to store this

information for future utilisation, it is not surprising that privacy protection is

considered as one of the most major properties that a PIS should take into

consideration from the very beginning of the design process with many researchers

proposing specific guidelines or models that may be applied [4, 7, 20, 25, 34].

Finally, the design of PIS should not be the concern of software engineers only.

Software engineers have the necessary skills to analyse and design a system from an

IS perspective: create entity-relationship diagrams, data-flows, large databases, select

the most appropriate technical solution to the given problem, and so on. Therefore,

they, most probably, lack the necessary skills to apply their design solution to the

physical space on an effective manner, not to mention to propose the most effective

solution based on the design problem. As such, the design team should be enriched

with additional members that may involve types of people such as architects, or

internal decorators, each providing a different perspective to the design of PIS. These

people should be involved from the early stages of the design process in order to

counsel software engineers on how to exploit environments’ smart spaces. Likewise,

they may advise them on how and where to place pervasive artefacts in an

unobtrusive, aesthetical, and possibly invisible, to the system users’, manner. Finally,

they may recommend alternative layout propositions regarding the effective

exploitation of the physical space in order to minimise hardware placement and

environment restoration costs.

4 Conclusions

This position paper outlined the components of a design theory for the development

of PIS. The theory consists of a set of meta-requirements, a set of meta-design

considerations, and a set of design method considerations. This research stemmed

from the lack of a consolidated framework to guide the design process of PIS. Our

intention in this paper is to present the constituting components of the proposed

design theory. As such, activities related to the validation of the design theory are

beyond the scope of this position paper, nevertheless we will briefly discuss them in

the following paragraphs.

Because ISDTs propose theoretical contributions [45, 46] we need to empirically

test their propositions. Only the accumulated weight of empirical evidence will, in

essence, establish the validity of any design theory without any doubt. In our case, the

proposed design theory has been employed for the design of a Pervasive Retail

Information System (PRIS). Specifically, we followed the theory’s prescriptions to

design and implement a PIS capable of enhancing the shopping experience in

supermarkets. The detailed design of the pervasive system has been published in [39].

To validate the propositions of the design theory we have generated a set of validation

hypotheses measuring the instance’s value and acceptance. To assess them, we

organised a field experiment in a Greek supermarket (ATLANTIK) where the PRIS

was used by supermarket shoppers to conduct part of their shopping. All research

hypotheses have been validating which indicates that the proposed design theory

results to valuable and acceptable PIS instances. The field experiment results

concerning the system’s value have been published in [23]. The field experiment

results concerning the system’s acceptance are available at [24].

The value of this research will be determined by the application of the proposed

design theory by other scholars, in their effort to design PIS instances. We believe

that this research provides an aggregated approach to describe PIS characteristics and

prescribe their development. Moreover, it represents the only consolidated approach,

to our knowledge, that investigates the problem of PIS design. In any case, the

proposed design theory aims at specifying generic design principles that should be

inherited by PIS instantiations. We decided to follow that paradigm in order to ensure

that the proposed design prescriptions are applicable to all PIS instances, irrespective

of their application domain. It is up to the PIS designers to interpret and apply these

prescriptions based on their design problem.

References

1. Abowd, G.D., Atkeson, C.G., Hong, J., Long, S., Kooper, R. and Pinkerton, M.

Cyberguide: A Mobile Context-aware Tour Guide. Wireless Networks, 3 (5) (1997) 421-

433.

2. Abowd, G.D. and Mynatt, E.D. Charting Past, Present, and Future Research in Ubiquitous

Computing. ACM Transactions on Computer-Human Interaction, 7 (1) (2000) 29-58.

3. Banavar, G. and Bernstein, A. Software Infrastructure and Design Challenges for

Ubiquitous Computing Applications. Communications of the ACM, 45 (12) (2002) 92-96.

4. Beckwith, R. Designing for Ubiquity: The Perception of Privacy. IEEE Pervasive

Computing, 2 (2) (2003) 40-46.

5. Bederson, B.B., Audio Augmented Reality: A Prototype Automated Tour Guide. in ACM

Conference on Human Factors in Computing (CHI '95), (New York), ACM Press (1995)

210-211.

6. Bellotti, F., Berta, R., De Gloria, A. and Margarone, M. User Testing a Hypermedia Tour

Guide. IEEE Pervasive Computing, 1 (2) (2001) 33-41.

7. Beresford, A.R. and Stajano, F. Location Privacy in Pervasive Computing. IEEE Pervasive

Computing, 2 (1) (2003) 46-55.

8. Berhe, G., Brunie, L. and Pierson, J.M., Modeling Service-Based Multimedia Content

Adaptation in Pervasive Computing. in CF'04, (Ischia, Italy) (2004).

9. Birnbaum, J. Pervasive Information Systems. Communications of the ACM, 40 (2) (1997)

40-41.

10. Checkland, P. Systems Thinking, Systems Practice. John Wiley and Sons, Chichester, UK

(1981).

11. Davies, N., Cheverst, K., Mitchell, K. and Efrat, A. Using and Determining Location in a

Context-Sensitive Tour Guide: The GUIDE Experience. IEEE Computer, 34 (8) (2001)

35-41.

12. Dey, A.K. Understanding and Using Context. Personal and Ubiquitous Computing, 5

(2001) 4-7.

13. Djajadiningrat, T., S., W., Frens, J. and Overbeeke, K. Tangible Products: Redressing the

Balance Between Appearance and Action. Personal and Ubiquitous Computing, 8 (5)

(2004) 294-309.

14. Gajos, K. and Weld, D., SUPPLE: Automatically Generating User Interfaces. in Intelligent

User Interfaces (IUI) 2004, (Funcha, Portugal) (2004).

15. Giaglis, G.M. On the Integrated Design and Evaluation of Business Processes and

Information Systems. Communications of the AIS, 2 (5) (1999).

16. Hackos, J. and Redish, J. User and Task Analysis for Interface Design. John Wiley and

Sons, New York, NY (1998).

17. Han, R., Bhagwat, P., LaMaire, R., Mummert, T., Perret, V. and Rubas, J. Dynamic

Adaptation in an Image Transcoding Proxy for Mobile WWW Browsing. IEEE Personal

Communications, 5 (6) (1998) 8-17.

18. Hansmann, U., Merk, L., Nicklous, M.S. and Stober, T. Pervasive Computing: The Mobile

World. Springer-Verlag, Berlin, Heidelberg (2003).

19. Held, A., Buchholz, S. and Schill, A., Modeling of Context Information for Pervasive

Computing Applications. in 6th World Multiconference on Systemics, Cybernetics and

Informatics (SC12002), (Orlando, FL, USA) (2002).

20. Jacobs, A.R. and Abowd, G.D. A Framework for Comparing Perspectives on Privacy and

Pervasive Technologies. IEEE Pervasive Computing, 2 (3) (2003) 78-84.

21. Jonsson, M., Context Shadow: An Infrastructure for Context Aware Computing. in

Workshop on Artificial Intelligence in Mobile Systems (AIMS) in conjunction with ECAI

2002, (Lyon, France) (2002).

22. Kasper, G.M. A theory of Decision Support System design for user calibration. Information

Systems Research, 7 (2) (1996) 215-232.

23. Kourouthanassis, P. Can technology make shopping fun? ECR Journal, 3 (2) (2004) 37-

44.

24. Kourouthanassis, P. A Design Theory for Pervasive Information Systems Department of

Management Science and Technology, Athens University of Economics and Business,

Athens, Greece (2006).

25. Langheinrich, M., Privacy by Design – Principles of Privacy-Aware Ubiquitous Systems. in

Ubicomp 2001, (Berlin, Heidelberg), Springer-Verlag (2001) 273-291.

26. Lei, Z. and Georganas, N.D., Context-based Media Adaptation in Pervasive Computing. in

IEEE Canadian Conference on Electrical and Computer (CCECE'01), (Toronto, Canada)

(2001).

27. Limayem, M. A design methodology for embedding decision guidance into GDSS. Group

Decision and Negotiation, 5 (2) (1996) 143-164.

28. Markus, M.L., Majchrzak, A. and Gasser, L. A Design Theory for Systems that Support

Emergent Knowledge Processes. MIS Quarterly, 26 (3) (2002) 179-212.

29. Mathiassen, L., Munk-Madsen, A., Nielsen, P.A. and Sage, J. Object Oriented Analysis and

Design. Marko Publishing, Aalborg, Denmark (2000).

30. Mohan, R., Smith, J. and Li, C.S. Adapting Multimedia Internet Content For Universal

Access. IEEE Transactions on Multimedia (1999) 104-114.

31. Mumford, E. and Weir, M. Computer Systems in Work Design - The ETHICS Method.

Associated Business Press, London (1979).

32. Nielsen, J. Usability Engineering. Academic Press (1994).

33. Norman, D.A. The Invisible Computer: Why Good Products Can Fail, the Personal

Computer Is So Complex, and Information Appliances Are the Solution. MIT Press

(1999).

34. Palen, L. and Dourish, P. Unpacking “Privacy” for a Networked World. CHI Letters, 5

(2003) 129-136.

35. Ponnekanti, S.R., Lee, B., Fox, A., Hanrahan, P. and Winograd, T., I-Crafter: A Service

Framework for Ubiquitous Computing Evironments. in UBICOMP2001, (Georgia, Atlanta)

(2001) 56-75.

36. Preece, J., Rogers, Y. and H., S. Interaction Design: Beyond Human-Computer Interaction.

Wiley and Sons, New York (2002).

37. Redstrom, J. Designing Everyday Computational Things Department of Informatics,

Goteborg University, Goteborg, Sweden (2001) 244.

38. Roussos, G. Appliance Design for Pervasive Computing. IEEE Pervasive Computing, 2 (4)

(2003) 75-77.

39. Roussos, G., Spinellis, D., Kourouthanassis, P., Gryazin, E., Pryzbliski, M., Kalpogiannis,

G. and Giaglis, G.M., Systems Architecture for Pervasive Retail. in ACM SAC Track on E-

Commerce, (2003) 350-356.

40. Saha, D. and Mukherjee, A. Pervasive Computing: A Paradigm for the 21st Century. IEEE

Computer, March (2003) 25-31.

41. Satyanarayanan, M. Pervasive Computing: Visions and Challenges. IEEE Personal

Communications, August (2001) 10-17.

42. Simon, H.A. The Sciences of the Artificial. MIT Press (1996).

43. Stein, E.W. and Zwass, V. Actualising organisational memory with Information Systems.

Information Systems Research, 6 (2) (1995) 85-117.

44. Story, M.F., Mueller, J.L. and Mace, R.L. The Universal Design File: Designing for People

of All Ages and Abilities. The Center for Universal Design, NC State University (1998).

45. Walls, J.G., Widmeyer, G.R. and El Sawy, O.A. Assessing Information System Design

Theory In Perspective: How Useful Was Our 1992 Initial Rendition? Journal of

Information Technology Theory and Application, 6 (2) (2004) 43-58.

46. Walls, J.G., Widmeyer, G.R. and El Sawy, O.A. Building an Information System design

theory for vigilant Executive Information Systems. Information Systems Research, 3 (1)

(1992) 36-59.

47. Weaver, P.L., Lambrou, N. and Walkley, N. Practical SSADM+. Pitman, London (1998).

48. Weiser, M. The Computer of the 21st Century. Scientific American, 265 (3) (1991) 66-75.

49. Weiser, M. The Computer of the 21st Century. IEEE Pervasive Computing, 1 (1) (2002)

19-25.

50. Weiser, M. Some Computer Science Issues in Ubiquitous Computing. Communications of

the ACM, 36 (7) (1993) 75-84.