Some grammatical features are used much more often in some types of text, or genre, than in others. For instance, imperative clauses (like Chop the carrots finely; Beat the mixture until smooth) are common in instructional genres such as recipes – for obvious reasons.

However, sometimes the reasons for using a particular grammatical structure are less obvious. For instance, why does a speaker or writer use the passive voice (as in The house was sold or The house was sold by his sister) rather than the active (His sister sold the house)? There may be different reasons in different contexts.

In this investigation we are going to look at the passive, and whether it is used more often in some genres than in others. This may help us in thinking about why the passive is used.

Spoken vs. written English

The broadest genre distinction is between spoken and written English. We might pose the question:

  • Is the passive used more often in written than in spoken English?

Spoken vs. written: Step 1

Step 1. We start by looking for main clauses which are passive in ICE-GB. Our search finds the following:

  passive main clauses
Spoken: 2,241
Written: 3,321
Total: 5,562
  • It looks as though there are around 50% more passive clauses in the written material than in the spoken. But is this true?

We can’t directly compare the numbers for spoken and written numbers of passives. Why?

  • The corpus contains different amounts of spoken and written material (about 60% is spoken and 40% written).
  • Main clauses in speech tend to be short and snappy compared to those in writing. This means:
    • We expect to see more main clauses per thousand words in spoken transcripts than in written text.
    • So there are more opportunities to use a passive main clause in speech.

To do a proper comparison, we need to:

  • find out the total number of main clauses in the spoken and written material, and then
  • work out what proportion of these are passive.

Spoken vs. written: Step 2

Step 2. A search for all main clauses in ICE-GB finds the following:

  passives main clauses
Spoken: 2,241 45,334
Written: 3,321 23,722
Total: 5,562 69,056

Spoken vs. written: Step 3

Step 3. Divide the number of passive main clauses by the number of main clauses to answer the question:

  • What percentage of main clauses are passive?
  passives main clauses proportion
Spoken: 2,241 45,334 5%
Written: 3,321 23,722 14%
Total: 5,562 69,056 8%
  • Now we can see that a much higher percentage of clauses are passive in the written material than in the spoken material. 
  • But instead of there being 50% more (as it seemed at Step 1), it’s almost three times as much!

Comparing written genres

At a more detailed level, we can compare use of the passive in different types of written genre.

There are six different kinds of printed written material contained in the corpus (we will leave aside the non-printed written material, such as letters):

  • academic writing
  • creative writing (novels and stories)
  • instructional writing (administrative/regulatory and skills/hobbies)
  • non-academic writing
  • persuasive writing (press editorials)
  • reportage (press news reports)

Which genre do you think will contain the highest proportion of passives?

...And the lowest proportion of passives?

We will again search for passive main clauses and all main clauses, and then calculate the proportion of the main clauses that are passive.

This time we will investigate the different printed written categories.

Comparing written genres: results

We find the following:

  passives main clauses proportion
Academic: 907 3,967 23%
Creative: 160 3,510 5%
Instructional: 464 2,362 20%
Non-academic: 772 4,580 17%
Persuasive: 122 1,066 11%
Reportage: 307 2,392 13%
All printed: 2,732 17,877 15%


  • We can see that the proportions of passive clauses vary considerably among the six genres.
    • Which genres have the highest and lowest proportions?
    • Were your predictions confirmed, or did you find something different?
  • Can you think of possible explanations for the patterns we find in the corpus?
  • Are there ways in which these possible explanations could be tested?

Examining a written extract

One way of exploring further is to look at one or more extracts in detail, examining examples of passives. For instance, we could choose one of the 40 extracts of academic writing in the corpus.

These individual extracts themselves differ in the proportions of passives used, so we could choose one with a particularly high proportion of passives.

Let’s look at some examples of passives taken from an extract about computer software design. The proportion of passives in this extract is 44% – much higher than the proportion for academic writing overall.

The passive verb phrases are highlighted so you can easily find them. (Remember that we searched only for main clauses which were passive. You may notice some other passive verb phrases which are not highlighted because they are in subordinate clauses, e.g. This paper proposes a means by which Mascot can be used ... .)

Mapping a Mascot 3 design into Occam

by Dick Knowles

Mascot is the MOD preferred software design method for embedded software systems.It has been widely used, both with and without toolsets, for the design of software written in Pascal and CORAL 66.A report of a study funded by RSRE proposed how to implement Mascot 3 systems using Ada.This paper builds on that study to propose a mapping to allow Mascot 3 to be used with Occam.The proposed method is easy to use without a toolset and appears to produce efficient Occam programs.



Mascot (Modular Approach to Software Construction Operation and Test) is a proven design method for large embedded software systems.It is the MOD standard design method for these systems.Mascot 2 has been in use since 1976 and a number of toolsets have been developed to support it.In 1987 the Mascot 3 handbook was published [1], and toolsets are now being developed to support Mascot 3 with Ada.

The transputer has many benefits for embedded systems.It offers considerable computing power for a small package size and is designed to allow simple physical design of multiprocessor systems.Ada is now available for transputers, but no toolsets are available and Ada ’s support for multiprocessor systems is minimal.This paper proposes a means by which Mascot can be used to develop Occam systems.


Alternative approaches


Use of a Mascot kernel

The initial approach explored was to develop a Mascot kernel for the transputer and to use it in some benchmark applications.This kernel included the full set of Mascot 3 primitives and a subset of control and monitoring.It was used in Occam versions of the two benchmark systems developed during the Mascot 3 with Ada study [2].

The performance of the benchmarks was promising, but other results were poor.The most serious problem was concerned not with the kernel itself but with the building of systems from their component Mascot units.

Mascot 3 separates interface specification modules from both modules that require them and modules that provide them.Interfaces contain procedure, function and possibly data declarations.Activity modules, which are scheduled by a kernel, require access interfaces in ports.Ida (Intercommunication data area) modules, which provide access interfaces in windows, manage the sharing of data between activities.The kernel provides synchronisation and exclusion primitives for Idas.Ports and windows are connected by paths.

Occam does not allow the separate compilation of a unit with multiple outer level procedure declarations.This means that Idas cannot be compiled separately.Systems therefore have to be built by copying the text of Idas and activities into single compilation units.

The mappings into Ada proposed by the Mascot 3 with Ada study required similar copying.The report therefore recommended development of toolsets to do this automatically.This could also be done with Occam, but the performance of the compilation system would suffer.Without a toolset, these problems would cause the programmer considerable difficulties and lead to errors.The reduction in compilation speed would also be particularly visible and annoying.

This problem is specific to Immos Occam and Pascal.It seems that other suppliers may be producing compilers that are suitable for separate compilation of Mascot Idas.The 3L Parallel C and Pascal compilers appear to be of this type.Such compilers will also support the mapping proposed by this paper, and the results below show that this may be preferable in some systems.


Use of Occam concurrent features

One of the changes between Mascot 2 and 3 is that Mascot 3 systems are not mandated to use the standard Mascot primitives.Instead, they allow ‘the (Mascot) model to be mapped onto equivalent features in a concurrent language’.This approach was therefore considered and found to be far more attractive.It is the approach proposed by this paper.

Occam expects separately compiled procedures to communicate via Occam channels.An Ida written as a separately compiled procedure can operate by accepting access procedure calls as messages from an Occam channel.Access interfaces can be mapped as Occam channel protocols.Individual access procedures can be mapped as variants of this protocol.With this mapping, the distinction between Idas and activities can still be retained, and this was found to be an efficient method of implementing Occam systems.


Benchmark results

Table 1 shows the performance of the benchmarks using the proposed Occam mapping.It compares these results figure 1 with the Occam kernel mapping and with results extracted from the Mascot 3 with Ada report.

These results show that the proposed transputer/Occam combination can be expected to outperform many other combinations, although the ratio will vary by application.The mapping method proposed is at least partly responsible for this performance.Unconstrained use of Occam tended to lead to less efficient solutions.

A number of comments should be made about the results.

• First, the hardware used for the transputer results was a T414 running at 15 MHz with 150 nS memory.With faster transputers, the results could be improved by at least a factor of 2.

• Second, the very small program sizes reflect the transputer ’s ability to multi-task without any runtime executive.The large program sizes of the other systems show how the application code size can be overwhelmed by the runtime system.

• The results for the Occam kernel mapping show an increase in code size to include the kernel.The times are typical of a range of kernels that were implemented with different implementation strategies and complexity.

With the proposed mapping, there is no kernel to support Mascot control and monitoring.If these facilities are required, an alternative approach would have to be used.However, the Mascot facilities that allow reusable modules to be combined into a variety of module and subsystem test harnesses are fully supported.

The results for the other systems may have been improved by versions of compilers released since Reference 2 was published.


Mapping rules

This Section describes the proposed rules for creating an Occam program from a Mascot 3 design.It is addressed to readers who have some knowledge of Mascot 2 or 3 and Occam.It is assumed that the Mascot 3 design has already progressed to the stage of graphical and textual representations described in the official handbook.To prevent confusion between Occam channels and Mascot Channels, all text referring to Mascot Channels will use a capital C.The basic rules are that □ each Mascot 3 template unit is mapped as a separately compiled Occam procedure; □all Mascot 3 definition and interface units are mapped into one or more Occam libraries; □each Mascot 3 path is mapped as one or two Occam channels.In parameters pass through one channel;any out parameters pass through the optional channel.Dummy parameters have to be added in some cases; □access interfaces are mapped as Occam protocols.A variant protocol is used for interfaces with more than one access procedure; □each access procedure call from an activity is mapped as a channel write of all in parameters, followed by a read of any out or synchronisation parameters.In cases where there is no parameter, dummy parameters are introduced; □each Ida is mapped as a main loop reading in a single ALT statement from all channels.

It is not possible to give a definitive set of rules in this paper.Instead, two examples are given showing how the mapping is achieved.


Simple Mascot Channel example

This first example is a subset of the program that was used as the Channel benchmark.It uses the traditional Mascot buffered Channel in a test network.

The Mascot 3 network is shown in Fig. 1.The Mascot 3 and Occam versions of the definition and access interface units are shown in Figs. 2 and 3.The Occam Figures use the conventions of the Inmos folding editor.The lines starting ... represent hidden folds containing additional lines of text.The sections bounded by {} symbols show the contents of such folds.The comment on the ... line is the same as that on the} line.

The absence from Occam of records and user-defined types causes difficulties when mapping definition units.In this example, objects of the position type have been mapped as arrays.The Occam version of the definition unit defines constants, with the same names as the record fields, that can be used to index position arrays.

Access interface get is mapped as two Occam protocols:one is used to send a dummy Boolean parameter to the Channel to request a record;the other is used to return the record.Variant protocols are not required because there is only one procedure in the access interface.

The Mascot and Occam versions of the Channel are shown in Figs. 4 and 5.The Occam version shows the standard structure for the Occam mapping of an Ida.The library defining the access interfaces has been USEd.The windows provided by the unit are declared as channel parameters of the procedure(the put access interface in this example does not need an out channel).These declarations use the Mascot 3 window and access procedure names suffixed with. in or. out to distinguish the two channels.

The body of the procedure is a loop, reading from all input channels with a single ALT statement.Fig. 6 shows the contents of the hidden gw window fold of the previous Figure.

In this example, because there is no variant protocol, guards can be used on the input statement.In the normal case, synchronisation is achieved by delaying the response on the output channel of the interface (possibly passing a dummy response parameter).Idas must be designed so that access procedure calls from different windows are accepted while responses are delayed.

The Mascot design will show the concurrency required by the external hardware connections and that for potential division of the solution across multiple processors.This concurrency is achieved by Occam PAR statements in the Mascot composite modules.Additional PAR statements are likely to lead to inefficiency.

Where access procedures are used by more than one window, common procedures can be declared to be called from the ALT statements.

If an access interface is called by more than one Mascot unit, an array of Occam channels must be declared.Each channel can then be dedicated to a single Occam process pair.

In this version of the Mascot Channel, the guard test consists of a check on the number of messages in the buffer.This was found to be faster than the normal comparison of pointers.It is not possible in normal Mascot implementations to use this method because the count variable cannot be safely updated from both reading and writing access procedures.Writing of Idas is simplified in Occam, since all accesses to the Ida operate in the same process and all data are therefore implicitly coherent.

The Mascot 3 and Occam forms of the test network description are shown in Figs. 7 and 8.The Occam can be derived from either the textual or graphical form of Mascot 3.

The used templates are declared by attaching them to their Occam declarations,(in some cases, such as Mascot library templates, they could be placed in an Occam library and USEd).

Unlike the Mascot 3 design language, it is necessary in Occam to declare all paths explicitly.These paths must then be connected to both their ports and windows.The definition and access interface declarations of the channels are accessed by the USE statements.

The instantiations of the templates are mapped as procedure calls passing the constants and path connections as parameters.


Track pool example

This example is a subset of the program that was used for the pool benchmark.It contains a simple pool Ida that manages aircraft track entries.It is required to allow updating of track entries by data from several sensors using a read and lock procedure call (prior to writing) and a write and unlock procedure call (to complete writing).Read only access by other windows must be allowed while a track entry is locked.Its graphical Mascot form is shown in Fig. 9.

Questions for discussion

  • What do these examples suggest about the use of the passive?
  • You might try rewriting some of them in the active voice for comparison.
  • Are there any difficulties in doing this?
  • Why do you think so many passives are used in this text?
  • These examples are all from an article by one writer. To what extent can we generalise from this?

You could further test your ideas by comparing another extract from an academic text. Choose one of the extracts displayed below. They come from:

  • a book on housing in an ageing society (26% passives) [W2A-013]
  • an article on accidents and safety regulations (33% passives) [W2A-018]
  • a scientific article about the atmosphere (54% passives) [W2A-029]



The first half of Chapter 5 looks at the interface between institutional provision (residential homes and nursing homes) and housing provision.The second half of this chapter looks at the interface between community care and housing policy.The future development of government policy towards both institutional care and community care will have significant repercussions upon the housing finance implications of an ageing population.

The demography section of Chapter 1 underlined the ageing of the elderly population.It is the ‘old old’ or those over 75 who are most likely to experience major health and mobility problems.It has been estimated that by 2001, more than 3 million people over 65 will be living alone, 185,000 will be unable to get out of bed unaided and nearly 700,000 will be suffering from dementia (Sinclair, 1988, p 246).Such figures lead some commentators to talk of a demographic time bomb.These trends are of major concern to a government with a strong policy objective to control public expenditure on all its social programmes (Thain and Wright, 1990) and to reduce the level of direct state involvement in service provision (Flynn, 1989).

This combination of factors has initiated a major debate about how best to meet the growing care need of the ‘old old’.Should this be through institutional care or should elderly people be encouraged to remain in their own homes?This debate has important implications for housing finance issues.For example:1.Would a growth of institutional care influence future inheritance patterns through residents paying their fees from home sale receipts?2.Should elderly people be encouraged to use home equity release schemes to meet their health and social care needs?3.What are the design and adaptation implications of encouraging older people with disabilities to remain in their own homes?

These and other issues are discussed in this section.First, however, it is necessary to say something about the changing politics of community care to put the subsequent discussion in context.

The changing politics of community care

In December 1986, the Secretary of State for Social Services asked Sir Roy Griffiths “to review the way in which public funds are used to support community care policy and to advise me on the options for action that would improve the use of these funds as a contribution to more effective community care” for what the report calls “adults who are mentally ill, mentally handicapped, elderly or physically disabled and similar groups” (Griffiths Report, 1988, p 1)

Community Care: Agenda for Action was published in March 1988 and called for social services departments to be given the lead role in community care planning at the local level for all these need groups; for central government to provide earmarked funds for service development; and for a mixed economy of provision in which the private and voluntary sectors would play the lead role in service provision.Local authority social services departments would plan and regulate; the private and voluntary sectors would provide services; and health authorities would concern themselves only with the medical components of community care.Sir Roy also argued for more public expenditure to be directed to community services and less towards funding the fees of elderly people in private residential homes and nursing homes.With regard to housing, he argued that housing authorities should only provide “bricks and mortar” so that, for example, all wardens of sheltered schemes for rent should be employed by social services authorities.The report also argued that better-off adults had a responsibility to plan for their own social and health care needs so that public resources could be targeted at those in greatest need.

The Griffiths Report was not received with great enthusiasm by the Conservative government who disliked the emphasis upon local authorities as lead agencies and the idea of a major expansion of earmarked funds.A White Paper entitled Caring for People: Community Care in the Next Decade and Beyond was, however, finally published in November 1989 (Secretaries of State, 1989).This confirmed the lead role of social services departments, emphasised the need to maximise the use of the private and ‘not for profit’ sectors as service providers and changed the arrangements for providing public subsidy for those in private residential and nursing home care.However, the White Paper offered little in the way of earmarked funds other than a limited scheme for those with mental health problems.The main proposals of the White Paper will come into operation in April 1991.

With regard to housing, the White Paper had much more to say about its importance than the Griffiths Report.“Suitable good quality housing” is seen as “essential” to social care packages (Secretaries of State, 1989, p 9) and the importance of improvement grants, Care and Repair schemes and Staying Put schemes are all stressed (Secretaries of State, 1989, p 25).It is argued that “social services authorities will need to work closely with housing authorities, housing associations and other providers of housing of all types in developing plans for a full and flexible range of housing” (Secretaries of State, 1989, p 25).

The groups covered by the Griffiths Report and the subsequent White Paper were neglected in the welfare legislation of the 1940 s.Services for these groups tended to be institution based, whether provided through the National Health Service (NHS) or local government, and they were often starved of funds relative to that available for the acute sick (in the NHS) or for children (in local authorities).There was a professional pessimism about the ability to help the so-called ‘chronic sick’ and so the neglect of their services seemed justified.A common term for the services of these groups became ‘the Cinderella services’, an attempt to express their failure to capture resources (Means and Smith, 1985).

The development of legislative support for domiciliary services and community care provision has been slow.The 1946 National Health Service Act enabled local authorities to provide a home help service.The provision of residential accommodation was the main responsibility placed on local authorities by the 1948 National Assistance Act.Section 29 did empower local authorities “to promote the welfare of persons who are blind, deaf or dumb, and others who are substantially and permanently handicapped by illness, injury or congenital deformity”.The Act did not allow local authorities to promote the welfare of elderly people, or to carry out preventive social work with this group.Local authorities could grant aid to a voluntary agency to provide a ‘meals on wheels’ service, but it was illegal to provide a direct service of its own.The voluntary sector, and especially local old people ’s welfare committees (now local branches of Age Concern) were encouraged to play the lead coordination role in visiting schemes, luncheon clubs, meals on wheels services and so forth.

Gradually, this situation changed.A 1962 Amendment Act to the 1948 legislation empowered local authorities to provide their own meals on wheels services.The 1968 Health Services and Public Health Act gave local authorities the general power to promote the welfare of elderly people, and the 1970 Chronically Sick and Disabled Persons Act placed a further set of obligations on local authorities, with regard to community care provision (for example, assessment for a telephone).But the implementation of these last two Acts was delayed until April 1971 so as to coincide with the creation of unified social services departments.These new departments were soon dominated by child care concerns, especially after the death of Maria Colwell in January 1973.Hopes of a major expansion of domiciliary services were further hit by the first tremors on the public expenditure front in the mid-1970s with the oil crisis.Increasingly, debate shifted away from the need for overall increases in social care expenditure towards arguments about the need for a shift of priorities between domiciliary and residential provision (Means, 1986).

Government interest in community care became far more marked in the early 1980s.They were becoming concerned about demographic projections on the likely growth in the number of older people.They wished to close down expensive long-stay hospital provision for other community care groups, especially those with learning difficulties and those with mental health problems.A succession of government and semi-official reports criticised the performance of community care services, especially over slowness in running down long-stay hospitals.This was seen as partly the result of organisational fragmentation between health authorities, social services authorities and housing authorities.These reports complained increasingly at how private residential homes and private nursing homes were being used to facilitate hospital run-down; consumers were being moved from one institution (NHS hospital) to another (a private sector home) rather than into ordinary housing, backed up by community services.The most damning report was produced by the Audit Commission (1986) entitled Making a Reality of Community Care.This report asked the government to establish a high level review and hence the appointment of Sir Roy Griffiths.

Issues in residential care

Perhaps the most striking feature of institutional care in Britain has been the consistent percentage of older people who enter such care over time.Over the last four centuries, this percentage has only varied from 3% to 6% (Fennell, Phillipson and Evers, 1988, p 29).Table 40 breaks down the different components of institutional care, although unfortunately it was not possible to separate elderly people from younger adults who are chronically ill or have physical disabilities. table 40

Public institutional provision (NHS beds and residential homes) will now be considered and this will be followed by a review of the private residential and nursing home sector.The emphasis throughout will be on funding and home equity release issues.The section concludes by considering the likely impact of the proposals of the White Paper upon future trends in residential care.

The NHS inherited large numbers of long stay hospitals where the ‘chronic sick’ were kept in terrible conditions.Many of these patients were suffering from frailty rather than ill-health.Increasingly, local authorities were encouraged to take over responsibility for this group leaving the NHS to concern itself with those elderly patients requiring either a rehabilitation programme after ill-health or long-term nursing care.However, despite the growth in the elderly population, geriatric hospital bed provision has been in decline in recent years.Increasingly, long-stay elderly patients are encouraged to enter private nursing homes.Elderly people in NHS beds retain their pension and the full costs of their care are met by the NHS.This is, therefore, the most expensive form of institutional provision by the state and one in which owner occupier patients are not under any pressure to use their home equity assets to meet their health care costs.We will see later that this changes very quickly once they enter a private nursing home.

Local authority residential homes were established by Section 21 of the 1948 National Assistance Act which stated that: it shall be the duty of every local authority .... to provide residential accommodation for persons who by reason of age, infirmity or any other circumstances are in need of care and attention which is not otherwise available to them.

The residents of these homes were expected to pay 21 shillings a week from their 26 shillings pension and keep five shillings for ‘pocket money’.This was in marked contrast with the previous system of public assistance institutions whose inmates lost their pension rights and where a charge for upkeep could be levied on other members of the family (Means and Smith, 1985).

In other respects the 1948 National Assistance Act had limited impact upon the previous system of institutional care.Residential homes were usually the same buildings as public assistance institutions and run by the same staff who were offered no retraining.It was not until the 1960s and 1970s that there was a major expansion of purpose-built residential homes.From the early 1960s, there was a growing critique of local authority residential homes as the prime social welfare provision for older people.Townsend (1964) in his national survey of homes argued that residents suffer from a loss of occupation;they fail to make new relationships;they experience a loss of privacy and identity;and they suffer a collapse of powers of self determination.Townsend (1964, p 190) concluded that residential homes failed to give residents “the advantages of living in a ‘normal’ community” and they should be abandoned.The alternative was to improve pensions, expand sheltered housing and develop preventive medical and domiciliary services.Over 20 years later Booth (1985, p 209) was complaining that all 175 homes in his sample: .... contain a very sizeable proportion of really quite independent residents, many of whom, according to staff, should not have been admitted in the first place or, after a spell of care, could have been moved into other accommodation more suited to their needs.

Accident Reporting Systems

Major disasters and dramatic incidents quickly hit the media, but these form only a tiny proportion of all accidents the HSE learns about.In Britain, any accident occurring at or in connection with work, and resulting in three or more days ’ incapacity, is reportable to the HSE.In addition, there are several specific categories of injury that are reportable regardless of how many days ’ incapacity result.These include certain eye injuries, electrical injuries requiring medical attention, loss of consciousness due to lack of oxygen, and any injury resulting in twenty-four hours or longer in hospital (RIDDOR, 1985).Additional considerations apply in particular areas:for example, the Railways (Notice of Accidents) Order 1986 applies to statutory railways, and virtually all reported railway accidents are reported under this Order.

In this way a very wide net is cast.A rich source of information on accidents occurring is available to the inspectorates of the HSE.This is in sharp contrast with the practice of the Occupational Health and Safety Administration (OSHA) in the USA, where only fatalities and catastrophes (FATCATS) are reportable to OSHA area offices.(Catastrophes are incidents involving five or more victims admitted to hospital.)

Up to March 1983, the bulk of accident reports in Britain came to the inspectorates indirectly.By agreement with the DHSS, copies of any forms claiming sickness benefit that appeared to arise from an industrial accident (or illness) were, in due course, forwarded to an HSE area office.The forms (‘B176s’) provided a sketchy description of the accident and characteristics of the victim.In the Factory and Agricultural Inspectorates, computer codes for type and agency of accident, type of injury, etc., were derived from these by clerical staff and stored on the computer system known as ‘SHIELD’.The forms were also scanned by an inspector (usually, but not always, a senior inspector) and a decision made whether or not to send an inspector to investigate.In addition to this roundabout (and now defunct) system, major accidents and fatalities were and are directly reportable by employers to the Inspectorate as soon as possible by telephone, with a follow-up written report.This system officially ceased at the end of March 1983.Under new regulations, which came into effect in 1986, all reports are to be made directly to the relevant enforcing agency within seven days, but major accidents must be reported by telephone as soon as possible.

Railway accidents have always been reported directly to the Railway Inspectorate.Certain particularly serious categories of accident are reported directly to Inspecting Officers by telephone.These include any accidents to passenger trains involving fatal or serious injury to passengers or train crews and any fatal accident at a level crossing.In addition, any particularly serious accident involving a freight train and any incident causing major disruption to a main line would also be reported directly by telephone.The only fatalities to the public not to be reported immediately are suicides and trespassers to the railway.In contrast to the Factory Inspectorate, fatalities to railway staff need not be reported by telephone unless a single incident has resulted in multiple fatalities.Other more minor accidents involving railway employees and passengers are reported directly to the Inspectorate on a monthly basis.

It is worth noting that important filters operate before an inspector receives notification that an accident has occurred, especially in relation to less serious injuries.Someone has already defined the incident as a notifiable accident, and constructed the available information into some kind of brief report.On the basis of this already filtered and framed information the inspector takes a decision whether to respond with an investigation.The process of defining what has happened and how to respond is well advanced before any investigation is carried out.

Although there are some variations between inspectorates of the HSE in the regulations applying to notification of accidents that fall within their authority, much greater differences arise in the ways in which the inspectorates respond to the reports they receive.The Mines and Quarries Inspectorate, for example, investigates only major accidents, and it investigates all of them.The Factory, Agricultural, and Railway Inspectorates investigate across the whole range of reported accidents.Out of a sample of 232 accidents investigated by the FI in 1982-3, 73 per cent were non-major, reported through the indirect B176 system via the DHSS, while 26 per cent were reported directly as major accidents (see also Table 2 above).Before we discuss the functions of investigations, we will look briefly at how the FI makes the selection of accidents for investigation.This selection implies that a visit (or possibly several visits) will be made by an inspector, and a written report made of what is found and what action is taken.Accident reports may be followed up in other ways -for example, a telephone call may be made to the establishment to clarify what happened;the establishment may be asked to forward its own report to the Factory Inspectorate; or a note may be left on the establishment file for follow-up at the next visit.

The Decision to Investigate

While the B176 system was operating, approximately 4 per cent of accidents reported to the FI were selected for investigation.Written guidelines for factory inspectors provide a basic list of the various factors to be taken into account in making this selection.For some categories of accident the guidelines specify that investigation is mandatory.These include fatalities, very serious injuries, and multiple casualties.When investigation is optional the factors that according to the guidelines, may influence the decision to investigate include whether a serious breach of the law is indicated; recurring accidents at a particular factory; and accidents to young people or children.

Empirical data from our research confirm that the factors listed in the guidelines do indeed affect inspectors ’ decisions to investigate.Comparison between accidents selected and accidents rejected reveal clear patterns of choice, reflecting the seriousness of the injury, and whether a breach of regulations is apparent.These patterns were confirmed by interviews with inspectors about how they make the selection, and by getting inspectors to talk aloud as they actually made the selection.In addition, certain types of accidents (such as explosions) or accidents at certain classic dangerous machines (such as horizontal milling machines), tend to be investigated.A major consideration is whether or not the inspector anticipates that some action can be taken as a result of the investigation.For this reason, quite serious injuries may not be investigated.For example, finger amputations are frequently not investigated.When asked why, inspectors said that there is usually little action an inspector can take following a finger amputation on woodworking machinery (except perhaps where the victim is a young trainee).Similarly, a broken limb caused by slipping in a car park is unlikely to be investigated by the Factory Inspectorate.

As a rule inspectors reject for investigation accidents where little can be done:but significant exceptions are those accidents where there is a public expectation that the inspectorate will respond.Lengthy investigations often follow fatal accidents even if it was apparent early on that there was little the inspector could do as the victim had made a fatal mistake.For the Agricultural Inspectorate in particular, accident reports can also be an important source of information about new agricultural machinery, especially imported machinery.Accident reports may be the first way that Agricultural Inspectors are made aware of the presence of hazardous machinery, possibly failing to comply with safety standards.If accidents appear to involve newly introduced machinery that fact may therefore be a reason for investigating.

Similar factors determine which accidents the Railway Inspectorate investigate, although in the case of this Inspectorate the situation is complicated by the various types of investigation available.Railway Inspectors can investigate accidents either under the 1871 Regulation of Railways Act or under the 1974 Health and Safety at Work Act.While most investigations are now undertaken under the latter, there is a strong preference to hold Inquiries under the 1871 legislation.This is partly a result of public expectation, but is principally because of the opportunities the 1871 legislation affords for publishing the findings of the investigations and disseminating information about accidents and their causes.The Health and Safety at Work Act, by comparison, imposes constraints on publication.

Unless the Secretary of State decides to appoint a member of the legal profession to conduct a formal Inquiry, the decision to hold an Inquiry is made at RI headquarters, either after an inspector has held a preliminary investigation or after the railway have conducted their own initial inquiry and provided a report.Account is taken of such factors as whether or not there was either actual serious injury or death or the potential for these; the degree of public interest in the accident; whether or not passenger trains are involved; whether or not the causes of the accident are known; how unusual the circumstances of the accident are; and whether or not non-compliance is suspected.If the accident is especially worrying in any of these respects, it may be decided to hold a public hearing of the evidence under the 1871 legislation (see below).


The decision to investigate an accident almost always implies that a visit by an inspector will be made to carry out an on-site investigation.The inspector making the visit could be anything from a trainee to a principal inspector.(Indeed, the accident may be selected for investigation largely because of the value to trainees of investigating a range of accidents.)A handful of less serious accident investigations fall by the wayside before a visit is made because of the time that has elapsed.For all the inspectorates we studied, ‘accident investigation’ is a class of visit in itself.However, an investigation is often combined with other activities.Thus, an inspector may use the same visit to check on other matters already known about, or to conduct a full basic inspection.About 5 per cent of Factory Inspectorate investigations bring the inspector to premises not previously registered with the HSE.Investigating the accident may then be secondary to impressing on the employer the requirement to register, and finding out more about the previously unknown and uninspected establishment that has been brought to light.

The investigation itself obviously takes varying lengths of time and may be conducted in a variety of ways, depending on the particular accident.It would, of course, be usual for the inspector to look at the site of the accident and interview any witnesses or other members of the workforce or management with relevant information (for example, about usual work practices; safety measures taken before and since the accident; other similar machines in use).As a general rule, the more serious the accident the more time will be devoted to its investigation.Following a fatal accident the inspector will be making a report to the Coroner.In addition, there will be a strong expectation that the inspectorate will make a thorough investigation and it will need to be seen to do so, even if it is clear from a fairly early stage in the investigation that there is little more to be learnt.

Closely linked with the seriousness of the accident is the action that the inspector anticipates taking.If it appears from the initial accident notification that a breach of regulations has resulted in a serious injury there is a strong possibility that prosecution will be considered (see further below).Once an inspector anticipates a possible prosecution, his or her approach to investigation necessarily becomes more formal.Statements need to be taken, and evidence assembled with an eye to the possibility that it may be presented in court.The investigation is not only more formal but also more time-consuming, and the urgency of collecting evidence before it is lost may mean that the investigation is carried out as a priority.

Once an inspector has concluded the investigation, a written report will be prepared outlining the findings of the investigation, any action taken, and any further action intended or recommended (such as prosecution).The investigation report itself may simply disappear into a filing cabinet.However, the information gathered at investigations may be directed into a number of channels feeding into policy at some level.A very effective use of accident information can be through National Industry Groups (NIGs) within the Factory Inspectorate.NIGs are part of the organization of Area Offices and staffed by field-level inspectors.Each NIG keeps an eye on a particular industry - such as chemicals, engineering, steel - providing a national focus.If it appears that an accident has wider implications that should be communicated to the industry or to inspectors dealing with the industry, the relevant NIG provides a channel for doing so.For example, in the course of our study an accident occurred at a wire-drawing machine, wrapping wire many times around a worker ’s hand causing such serious injury that the hand had to be amputated.Such an accident had not, in the inspector ’s experience, occurred before at that type of machine.Details of the accident and investigation were therefore passed to the wire rope and cable NIG.

A study of aerosol properties and solar radiation during a straw-burning episode using aircraft and satellite measurements


Meteorological Research Flight, Royal Aerospace Establishment, Farnborough

(Received 5 December 1988, revised 28 February 1990)


Aircraft measurements of radiative fluxes over the North Sea during a heavy straw-burning pollution episode are presented and compared with results from a radiative transfer model.

The model uses aerosol properties derived from aircraft and satellite measurements, and acceptable agreement with the measured fluxes is obtained.

High levels of aerosol concentration are described, with correspondingly large radiative effects.Boundary layer heating rates of 3 K day -1 are measured and rates of 5-1O K day -1 OYMENU in regions of higher aerosol loading are inferred.


In recent years there has been considerable interest in the effect of aerosol on solar radiation in the cloud-free atmosphere, much of it directed at influences on global and regional climate.Concern over possible global warming or cooling by anthropogenic aerosols has been heightened, especially by current work on the aftermath of a nuclear exchange as well as the effects of widespread deforestation.Study of the behaviour of radiation and aerosol tends to have been channelled into three areas, with limited interaction; measurement and collation of aerosol properties, modelling of the atmospheric radiation field, and measurement of the radiation field.

A large amount of information is available on the physical and optical properties of aerosol of different types and localities (Carlson and Benjamin 1980; Harris and Gerber 1982; Patterson et al. 1986; Kilsby and Smith 1987).However, owing to the extreme variability and range of concentration and optical properties of atmospheric aerosol, it is very difficult to model the radiation field reliably from climatological or standard aerosol properties.

Modelling of local and global climatic effects has been carried out using measured and standard aerosol properties (Braslau and Dave 1973; Wendling et al. 1985; Blanchet et al. 1986).Much uncertainty remains however, and a range of aerosol single scattering albedos exists at which under certain conditions, a line dividing global cooling and warming may be drawn (Charlock and Sellers 1980).

The ultimate test of such modelling is comparison with measurements of the real radiation field;unfortunately these are difficult measurements to make with sufficient accuracy.Radiation profiles allowing flux divergences to be inferred have been measured (Roach 1961; De Luisi et al. 1976; Method and Carlson 1982; McArthur 1984; Kitchen and Squires 1984), but seldom with sufficient accuracy or detail to stand comparison with profiles modelled using simultaneous aerosol measurements.

This work presents results of an experimental and theoretical study of a highly polluted atmosphere.Aircraft and satellite measurements of properties of aerosol resulting from straw-burning are combined in a radiative transfer model to be compared with measured profiles of downward, downward diffuse and upward radiative fluxes.Comparisons are made for the total solar band (0.3-3 μ6;6;6;), and the visible (0.3-0.7 μ6;6;6;6;) and near infrared (0.7-3 μ6;6;6;6;6;) parts of the solar band and good agreement is obtained throughout.It is found that although the aerosol has its origin in combustion, it is only a moderate absorber, in contrast with the very strong absorption assumed in some “nuclear winter” scenarios.These scenarios assume the production of absorbing aerosol from combustion of material in urban environments rather than combustion of vegetation.Combustion of vegetation undoubtedly forms a substantial part of any nuclear winter scenario, and additionally occurs with increasing frequency in slash and burn agriculture and deforestation world wide.In this paper, a severe pollution event is described, and boundary layer heating rates of 5-10 K day -1 are calculated.

The comparison of satellite-derived aerosol optical depths with aircraft measurements is particularly important, since validation of the satellite technique is essential before it can be used to provide much needed global and temporal information on aerosol.This information can be used to correct satellite-measured surface parameters (e.g. vegetation indices and albedo) as well as being directly included in climate models.


(a) Conditions

On 11 and 12 September 1985 the centre of an anticyclone was situated over Denmark and produced clear skies over the North Sea (Fig. 1).Following a fairly wet summer, a dry spell in early September gave farmers a good opportunity to burn straw and stubble.Straw-burning fires were widespread over the cereal-growing areas of England, and imagery from NOAA-9 Advanced Very High Resolution Radiometer (AVHRR) channel 3 (section 2 (c)) shows some 750 separate fires (≈ 1 km 2 resolution) each consisting of a blazing field (Fig. 2).

The south to south-westerly surface flow carried aerosol from the fires over the North Sea, trapped under two temperature inversions at 920 mb and 1000 mb (Fig. 3).Aerosol concentrations were particularly high near to the east coast of England, with large amounts of pollution also in the eastern North Sea, probably originating in the industrial regions of central Europe.

Since the current day ’s smoke plumes appear to have reached only some 70 km distance from the coast (section 3 (b)) the aerosol in the operating area (Fig. 1) on 12 September must be assumed to be the product of earlier burning on 11 September.Comparison of the aircraft profiles with radiosonde ascents from Hemsby at 0000 GMT and 1200 GMT that the lowest inversion (at 1000 mb) is consistent with that produced by overnight cooling, sustained by the cool (≈ 13 ºC) sea surface during its transport to the operating area.This process resulted in very high concentrations of smoke in the lowest layer, whereas over land and nearer the coast the aerosol was mixed throughout a much deeper boundary layer (and so was more dilute).

(b) Aircraft flight pattern and instrumentation

On 12 September the Hercules of the Meteorological Research flight (MRF) performed a sortie around 56ºN, 0ºW (between 1050 and 1445 GMT) -the measurement runs of the flight pattern are shown in Fig. 1.In addition a series of air sampling runs were performed off the north east coast of England on 11 September.The Hercules is equipped with meteorological and navigational equipment and other specialised instrumentation (Readings 1985), of which on this occasion the radiation and aerosol systems were used.The flight pattern consisted of a series of straight and level runs of 20 km length.Runs were performed at 12 levels between 60 and 8000m with interconnecting profiles for measuring temperature, dewpoint and aerosol properties.

Solar radiation measurements were made with four Eppley PSP pyranometers arranged in pairs facing upwards and downwards.One of each pair was fitted with a clear glass dome (WG295 transmitting between 0.3 and 3 μ6;6;6;) and the other with a red dome (RG715 transmitting between 0.7 and 3 μ6;6;6;6;).This arrangement allows measurements to be made in three spectral bands: solar, near infrared and (by subtraction) visible.Total upward and downward fluxes were measured as well as the diffuse component of the downward flux.The measurement procedure is described in section 4.

The aerosol scattering coefficient at 0.48 μ6;6;6; was measured continuously using a MRI model 1550 integrating nephelometer.This was calibrated before flight using Freon 12, Freon 22 and filtered air according to the procedure proposed by Ruby and Waggoner (1981).Number concentration and size spectra in the radius range 0.25-3.75 μ6;6;6; were measured using a PMS Forward Scattering Spectrometer Probe (FSSP).Aerosol samples were taken on 0.4 μ6;6;6; pore size Nuclepore membrane filters over 20-minute periods.These were analysed firstly for mass loading (± 20 μ6;6;6;) and secondly for absorption coefficient at 8 wavelengths between 0.5 and 1 μ6;6;6;6; using the Integrating Sandwich technique (Clarke 1982; Foot and Kilsby 1989).

(c) Satellite data

NOAA-9 (AVHRR) data from a pass overhead the United Kingdom at 1346 GMT on 12 September 1988 have been used for two purposes.Firstly the imagery from the near infrared channel 3 (3.7 μ6;6;6;) shows the distribution and extent of straw-burning over England (Fig. 2).Channel 3 is very sensitive to the presence of small hot spots in the field of view due to the non-linearity of the Planck function (Saunders 1986) and has been proposed as a means of monitoring straw-burning (Muirhead and Cracknell 1985), and used to monitor tropical deforestation in South America (Malingreau et al. 1989).

Radiances from the visible channel 1 (0.63 μ6;6;6;) and infrared channel 2 (0.86 μ6;6;6;6;) are also used quantitatively to derive aerosol optical depths.The technique is described in detail in section 3 (b).


(a) Aircraft measurements

The vertical distribution of aerosol is best shown by the total number concentration (0.25 &less-than;r&less-than;3.75 μ6;6;6;) measured with the FSSP (Fig. 4).This shows some aerosol situated below a temperature inversion at 900 m, but the majority is concentrated below a lower inversion at 90 m.No significant amounts of aerosol were observed above 900 m.

The vertical distribution of aerosol scattering coefficient, measured with the integrating nephelometer during aircraft profiles and level runs, is shown in Fig. 5.This corresponds very well with the FSSP measurements, but there is some discrepancy between the profile measurements, taken during a 2.5 ms -1 descent, and the level runs which were performed in an ascending staircase.This could be due to time lags in the instrument sampling and in flushing the sample volume of the instrument, especially after filling with the very polluted air of the lowest run of the ascent.The best estimate of scattering coefficient is considered to be between these measurements, although an underestimate may occur near the surface in the very high concentrations.

The size distribution by volume measured by the FSSP averaged over 12 minutes in the dense haze is shown in Fig. 6, and displays a dip at 0.9 μ6;6;6; radius characteristic of previous FSSP measurements in a variety of conditions (Kilsby and Smith 1987).This dip is believed to be instrumental in origin.The instrument is incapable of detecting the very small (r ‹ 0.2μ6;6 m) smoke particles expected from straw-burning, and high concentrations of large particles suggest that at a distance of 200-300 km from the source, some coagulation may have occurred.

A number of filter samples were collected on the 2 days of the experiment; three off the north east coast on 11 September at 60 and 300 m, and one on 12 September between 60 and 300 m over a 25-minute period.The results of gravimetric and optical absorption analysis of these filters are given in Table 1.A spectral plot of absorption coefficient (Β abs) is given in Fig. 7 for the 12 September filter.This displays the inverse wavelength dependence characteristic of a grey absorber such as elemental carbon.

Combination with mass loading measurements allows the specific absorption and effective imaginary refractive index (n 2) to be calculated (Gerber 1983).The specific absorption (Β a) of ≈ 2.2 m 2 g -1 is very high for normal atmospheric aerosols but is in line with accepted values for products of flaming combustion.Values in the literature range from 0.7 to 2.4 m 2 g -1 for burning pine needles (Patterson et al. 1986), through 2.0 m 2 g -1 for assumed nuclear winter smoke (NRC 1985) to 10-12 m 2 g -1 for carbon soots.

The single-scattering albedos (ω 0) at 0.55 μ6;6;6; have been calculated using the measured absorption coefficient and the scattering coefficient, measured at 0.48 μ6;6;6;6; scaled to 0.55 μ6;6;6;6;6;.The scaling factor of 1.16 was calculated from the wavelength dependence of the Standard Radiation Atmosphere (SRA) continental aerosol model (WMO 1986).The value of 0.92 (± 0.02) for 12 September is higher than those for 11 September but if a scattering coefficient for the profile is used rather than one for the level runs, a value of 0.90 is obtained.The relation between n 2 and ω0 is easily tested if assumptions of size distribution and n 1 (real part of the refractive index) are made.This is best illustrated by Fig. 8, in which the solid line is a mean relationship between n 2 and ω 0 calculated with Mie theory for a large number of observed size distributions with a refractive index, n 1, of 1.5 (Gerber 1983).The line is a good approximation to a number of measurements, and the straw-burning values are not exceptional.This comparison gives confidence in the three separate constituent measurements; absorption coefficient, mass loading and scattering coefficient.

Estimates of the total aerosol extinction optical depth on 12 September at 0.48 μ6;6;6; may be made using the nephelometer data shown in Fig. 5.The best estimate is 0.25 ± 0.04 assuming a single scattering albedo of 0.92, but this technique assumes horizontal homogeneity of aerosol concentration.A comparison with other estimates using different methods may be made using Fig. 10, and this vertical integration of the scattering coefficient appears somewhat low.

(b) Satellite measurements

The derivation of aerosol optical depth, and its spatial distribution over the ocean from satellites is a well established technique (Griggs 1984; Durkee et al. 1986; Masuda et al. 1988), though not without difficulty.In principle any visible or near infrared satellite channel may be utilised, and AVHRR channels 1 and 2 have been used here.AVHRR data are readily available and these channels give a useful compromise in spectral and spatial resolution, as well as being situated at nearly ideal wavelengths (0.58-0.68 and 0.72-1.1 μ6;6;6;) for accuracy of retrieval .


Englicious is totally free for everyone to use!

But in exchange, we ask that you register for an account on our site.

If you’ve already registered, you can log in straight away.

Since this is your first visit today, you can see this page by clicking the button below.


Englicious (C) Survey of English Usage, UCL, 2012-21 | Supported by the AHRC and EPSRC. | Privacy | Cookies