A reminder first of all of the way in which the STV would work in
parliamentary constituencies.
<p no=1>
Each constituency would return from three to nine members, five being
generally regarded as the optimum number.
<p no=2>
Every elector would have one vote.   It would not be cast by marking the
familiar X opposite the name of a single candidate.   Instead it would be
an expression of preferences, indicated by putting the figure 1 opposite
the name of the first-preferred candidate, the figure 2 opposite the name
of the second-preferred and so on.   The preferences could include as many
or as few of the candidates as the voter wished.
<p no=3>
A candidate would be elected not by securing a plurality of votes but by
securing a quota of votes.   A quota is established by dividing the total
number of valid votes by the number of members to be elected plus one, and
rounding up or adding one to the quotient.  In the framework of the STV
this apparently strange formula is rational.   In a typical United Kingdom
five-member constituency 250,000 votes might be cast.   The quota would
then be 250,000 divided by five-plus-one, i.e. 41,667.   No more than five
candidates could be credited with more than 41,667 votes each.   The quota
is the smallest number of votes to allow five candidates to be elected and
is just big enough to prevent the election of more than five.
<p no=4>
When the first-preference votes are counted any candidate who has secured
a quota is declared elected.
<p no=5>
After that - or if there are no such candidates - transfers come into
operation, entailing fresh counts.   Votes are successively transferred
from candidates who have a surplus of votes over and above the quota and
from candidates excluded as having the smallest number of votes in the
relevant count.  Votes are transferred to the candidate next in the
voter's order of preference who has neither been already elected or
excluded, i.e. to the next "continuing candidate", as the STV jargon has
it.
<p no=6 segment_break>
A company is regarded as dormant: in any period during which no
transaction occurs which must be recorded in the company's books, and if
it can claim small company exemptions under s.246, or could have but for
belonging to an ineligible group, and is not required to produce group
accounts.
<p no=7>
Advantage can be taken of the filing exemptions as a small company
provided that the appropriate note is included on the balance sheet.
<p no=8>
Disclose the names of persons who were directors at any time during the
year.
<p no=9>
Give details for directors at the year end of interests in shares and
debentures of any group company at the year end and at the beginning of
the year or subsequent date of appointment.
<p no=10>
Directors' report to be approved by the board of directors and signed on
behalf of the board by a director or the secretary of the company.   State
the name of the signatory.
<p no=11>
The balance sheet of the company must contain a statement from the
directors, immediately above the signature, that the company has been
dormant throughout the financial year.
<p no=12>
The balance sheet should be signed by a director on behalf of the board.
State the name of the signatory.
<p no=13>
Disclose the date on which the accounts were formally approved by the
board.
<p no=14>
A dormant company may pass a special resolution not to appoint auditors.
<p no=15 segment_break>
Shaped curtains fitted to arched windows can not be operational.  The heading
must be fixed to the arch, and the curtains are draped back and held open with
tie bands.  This treatment gives a lovely shape both day and night.  It is
also possible to use festoon or ruched blinds at this type of window, but they
would only draw up satisfactorily as far as the level at the bottom of the
arch.  For a very simple treatment, dress only the rectangular section below
the arch, leaving the top uncovered.
<p no=16>
 Doors that swing inwards present the problem of curtain fabric interfering
with the operation of the doors.  One of the most practical treatments with a
glass panelled door, but not necessarily the most attractive, is to anchor
sheer fabric to the top and bottom of the glass panel.  A roller-blind
covering the glass area is an alternative.  However, both these methods would
be greatly improved by adding full-length curtains at the sides where space
allows.  These could be hung from a decorative pole, perhaps set well above
and extending beyond the sides of the door, to enable the curtains to be drawn
well back.
<p no=17>
The choice of a suitable fabric can be daunting because there is such a wide
variety of fabrics on the market today.  The different combinations of colour,
beautiful designs, textures, fibres and weaves all add up to a vast range of
fabrics from which to choose and at enormously varying prices.  Ultimately,
however, the fabric you select for your window treatment should not only
complement your chosen style with its visual appearance but it must also suit
its purpose.  It would be inappropriate to pick a cotton repp, for example,
which is a thick, firm fabric for a feminine style with frills.  This style
requires a soft flowing material such as a lightweight cotton seersucker.
Conversely a loosely woven fabric would not be successful for a Roman blind,
which requires a firm, stable fabric.
<p no=18>
Apart from the visual appearance of curtain fabric, and suitability for its
purpose, the other important characteristic to consider is its drapability.
Soft, weighty fabrics tend to fall easily into gentle folds; an obvious
example is cotton velvet which is both soft, supple and heavy.  Every fabric
handles differently, and the only way to find out how it is likely to drape
when made up into curtains is to hold up a large sample and study how it
hangs.
<p no=19>
It is always advisable to buy furnishing fabrics rather than dress materials
as they come in wider widths and are usually of a more suitable weight.
Widths vary from 120cm(48in), 130cm(51in), 137cm(54in), 140cm(55in) and
150cm(59in).  Make a note of the width of your chosen fabric as you will need
this information when you come to estimate the amount of fabric needed.
<p no=20 segment_break>
An audit report is required for every set of accounts prepared for every
reference period (except dormant companies - s.253).
<p no=21>
The audit report should specify the standards which have been followed
during the audit.
<p no=22>
The audit report should state whether, in the opinion of the auditors, the
accounts have been properly prepared in accordance with the Companies Act
1985 and give a true and fair view of the state of the company's affairs
and of its profit or loss for the year.
<p no=23>
Reference should be made to the source and application of funds statement
in the opinion paragraph.
<p no=24>
If any of the following requirements have not been met, the auditors
report should state that fact:
<p no=25>
proper accounting records must be kept
<p no=26>
proper returns must be obtained from branches not visited by the auditor
<p no=27>
the accounts must be in agreement with the accounting records and returns
<p no=28>
all information and explanations considered necessary for the audit must
be given.
<p no=29>
If the information required on emoluments and other benefits of directors
is not given, the auditors must disclose it in their reports.
<p no=30 segment_break>
With c = 5.7 the flow is chaotic: the flow forms a single spiral embedded
in a disc, with trajectories from the outer part of the spiral twisted,
and folded back into the inner part of the spiral, forming a Mobius band.
The construction of the attractor is shown in Fig. 2.9.   This is similar
to the experimental attractor seen in Fig. 8.6.   A return map of a
section through the attractor looks like a single-humped, thickened,
one-dimensional map.   Three-dimensional views of the attractor are shown
in Fig. 2.10.
<p no=31>
As c is increased from 2 to 4.2, there is a sequence of period-doubling
bifurcations from a simple, period-one oscillation (Fig. 2.11).   Chaos
develops at the accumulation point of the period-doubling sequence, just
above c = 4.20, with families of similar orbits confined to thin bands
that grow from each of the period **f, orbits.   These attractor bands
are apparently separated by empty, repeller bands: a trajectory starting
between the bands is rapidly drawn into one of the attractor bands.
However, the attractor bands really form one, looped, attractor.   Pairs
of the attractor bands join in a sequence of reverse bifurcations until
the bands in the strange attractor meet [23,31].
<p no=32>
A further increase in the parameter c gives a change in the shape of the
attractor, which develops into a Rossler funnel as it expands with
increasing c.   The development and structure of the Rossler funnel are
illustrated in Figs 2.12 and 2.13.
<p no=33>
All the systems considered above contain a cross-term, in which the rate
of change of one variable is directly decreased by a term that is the
product of two other variables.   Such cross-terms can not occur in a
system that directly represents a chemical reaction, and the variables
represent concentrations: the concentration of one chemical species can
not be directly decreased by a process in which that chemical does not
take part [36].  Real chemical systems do show chaotic activity [5], but
this could be due to spatial, hydrodynamic irregularities rather than true
chemical chaos.
<p no=34 segment_break>
A system of equations that provides an accurate and physically
satisfactory representation of an experimental system can be cumbersome,
and even complicated and of high order, so solutions may often only be
obtained as numerical approximations to solutions.   Thus the numerical
solutions themselves may be considered to be approaching an equilibrium or
periodic solution.
<p no=35>
In the three-dimensional systems treated in section 2.2, the strange
attractors are locally planar: a small displacement perpendicular to this
sheet will decay, as the trajectory returns to the attractor; a small
displacement along the sheet will remain, as a trajectory is effectively
pushed forward in time; and a small lateral displacement will grow in
time.   This sensitivity to initial conditions (or instability for small
lateral displacements) is reflected by the single positive Lyapunov
exponent of a strange attractor of a three-dimensional system.   A strange
attractor of a four-dimensional system can have two positive Lyapunov
exponents, and so can have solutions that are more irregular than chaos.
<p no=36>
A simple system introduced by Rossler {;24-26}; that exhibits such
hyperchaos is: (2.6) **f Two-dimensional views of the strange attractor
are shown in Fig. 2.16: the xy plane view is reminiscent of the Rossler
funnel.   The system (2.6) has been obtained from a system similar to
(2.4) by adding a linear variable, w.   The variable z is activated
whenever a threshold value of x is exceeded, and the activation of z leads
to a reinjection of the trajectory to a new region in xyw space.  The
motion in xyz space is similar to that of (2.4): compare Fig. 2.17 with
Fig. 2.9b.   Three-dimensional outline views of the strange attractor in
xyz, xyw, zyw and zxw space are shown in Fig. 2.18.
<p no=37>
Although this is a very simple four-dimensional nonlinear system, it is
difficult to visualise its attractor.   One method of representing motion
in 4-space on a plane is by representing a point in 4-space by a line on
the plane, where each end of the line represents a pair of coordinates
[17]: this    is used in Fig.2.19 to illustrate the strange attractor of
Fig. 2.16.
<p no=38 segment_break>
Students in research areas that involved computing and physics appeared to
be particularly well-placed on the job market.
<p no=39>
The subject areas ( included in the above list) where no-one was looking
for employment were in physics, computing/physics, computing with non-IT
subjects, and in optical fibres.
<p no=40>
It is interesting to note that among those subject areas with lower than
average levels of students seeking employment were all the " purer" IT
specialisms.   There are two main interpretations that can be put on this
finding.  One appears logical but as yet is hypothetical, pending the
outcome of detailed discussions with the employers of IT specialists.
The other is of concern from a research viewpoint.
<p no=41>
The first hypothesis is that employers are familiar with single stream
research work, and in recruitment decisions, are acting conservatively.
So, research work closely related to developments in a single discipline
such as communications or micro-electronics are more easily
understandable, and may seem more applied and less esoteric than
cross-discipline work.
<p no=42>
The second interpretation is of more concern.  In completing the
questionnaire, research students may have opted to complete parts of it in
the quickest way without much thought.   There may not have been the
fullest consideration of the  range of IT subjects utilised in student's
research.
<p no=43 segment_break>
A shallow and wide window is not the best proportioned, and is often placed
high on a wall.  Privacy may not be a consideration, in which case the
simplest treatment may be a sheer to soften the window outline; or, for a more
cosy feeling at night, combine it with a roller blind.  If you want to improve
the window's proportions, use a cafe curtain and place the curtain rod just
above the sill level, allowing the curtain fabric(which should be opaque) to
cover the wall below to a suitable level.  (The window must be above eye-level
so that the sill is not visible.)  The curtain will create the illusion of a
deeper window.  You could add a smaller top tier of curtains which could be
drawn at night should you want complete privacy.
<p no=44>
Windows that pivot from a central point are frequently fitted in modern flats
for ease of cleaning.  The fitments will need to be placed well above and
beyond the sides of the window to give maximum freedom of movement.  Sheers
would present a problem at this type of window unless they are fixed onto the
top and bottom of the pivoting window-frame on rods or wire.
<p no=45>
 Dormer windows, fitted into alcoves in the roof, are difficult to treat
because there is little or no space into which the curtains can be drawn back.
Most blinds other than a roller blind will screen out too much light.  One
straightforward solution for sash windows is to allow minimum fullness in the
curtains themselves and tie them back tightly during daytime.  A small,
shallow valance approximately 15cm (6in) deep, with more fullness than the
curtains, would enhance the window's appearance and would make very little
difference to the light in the room.
<p no=46>
Another alternative would be to have cross-over sheers, which present a soft,
pretty appearance.  On inward-opening casement windows attach sheers, fitted
at the top and bottom on wire, and draw in the centre of each curtain with
ribbon tied into bows.
<p no=47 segment_break>
The notice of appeal sets out a number of grounds.   The first is that in
making a prohibited steps order prohibiting the parents from having any
contact with each other the justices erred in law as contact between
adults is not a step which could be taken by a parent in meeting a
parent's responsibility for a child and thus falls outside the terms of
section 8(1) of the Children Act 1989, by which a prohibited steps order
means an order that no step which could be taken by a parent in meeting
his parental responsibility for a child and which is of a kind specified
in the order shall be taken by any person without the consent of the
court.   
<p no=48>
Indeed, it is conceded by the representatives of the parties to this
appeal that an order made in the terms made by the justices was not open
to them in as much as it did not impinge upon the parental responsibility
of the parent concerned and impinged only upon contact between the parents
themselves.   
<p no=49>
To that extent that part of the order is plainly wrong and must be set
aside.   Of course, it is part and parcel of the whole decision of the
justices.   They considered that order necessary in conjunction with the
second part of their order relating to contact by the father with the
children.   
<p no=50>
To that extent it could well be said, and indeed I do find and hold, that
the effectiveness of the first part of the order takes away the substance,
as it were, of the whole order.   For that reason alone I feel able to say
that the decision of the justices was wrong and that therefore this court
is able to substitute its own discretion.   That is a ground which has not
been urged upon me by counsel in this case and I do not rely upon that
ground only in dealing with this matter.
<p no=51 segment_break>
 Bay windows present the same problem as corner windows: where do the curtains
hang during daylight hours?  One pair of curtains drawn back to the two outer
corners of the bay will look attractive, but they will cut down the light from
the side windows.  One solution for a large bay is to make one large pair of
curtains to fit across the front window and allow it to be free standing, that
is, do not fix its outer edges to the corners.  Make another pair of curtains,
one for each side window.  During the day, position the large pair at the
middle of the main window and draw in the centre of each curtain with tie
bands to form two attractive drapes.  Tie the side curtains back to the walls
adjacent to the side windows.
<p no=52>
Bay windows also lend themselves to blinds of all types.  Ruched or festoons
could add style to the setting.  Roman blinds would suit a squarish bay and
could be teamed with a pair of full-length draw curtains fitted across the
flat wall in front of the bay, or with dress curtains standing at the sides of
the bay.
<p no=53>
 Bow windows are a natural asset to any room and should be highlighted.  They
are not difficult to deal with, thanks to the many tracks on the market which
can be bent to fit the shape.  Floor-length curtains which hug the curve
present a beautiful flow of fabric, and can be drawn back to the sides of the
window to expose the sweeping curve during the daytime.  A valance following
the outline of the windows can offer another pleasing effect.  If there is a
window seat, curtains should hang to sill level.  Or a pair of full-length
curtains could be hung from a pole across the wall in front of the bow,
framing the windows by day and closing off the alcove at night.
<p no=54>
 Arched windows are one of the most elegant forms of window, and any treatment
which hides the beauty of the classical shape seems like sacrilege.  They do
pose difficulties in dressing, however, and in some instances there is no
alternative but to hang draw curtains from a track or rod placed above the
arch.
<p no=55 segment_break>
Redundancy of effort.  Scholars who are good at knowing where to look for
the sources of their subject matter may not be good at knowing where to
look for suitable software.   The replacement of mainframe use by desktop
computing in recent years means that scholars working alone do not have as
much contact with computer-learned people as they used to do.   This means
they are less in touch with software developments.   Hence, if they get
interested in designing their own software, scholars may spend their time
reinventing last year's wheel.
<p no=56>
Diversion of effort.  Even if the software designed is genuinely original
and useful, there is a danger that scholars become hooked on the
production of software.   There is a risk that a scholar whose talents and
academic position call for historical and critical work will spend most of
his or her working time as a programmer.
<p no=57>
Diversion of funding.  Because there is pressure on departments in
humanistic subjects to appear up to date and efficient, it is much easier
to persuade funding bodies to give money for computers and software than
to buy manuscripts, rare books, or second and third copies of frequently
used library texts.   After a few years a department may be left with
serious gaps in its library and a load of superannuated computing
equipment.
<p no=58>
Distortion of research.   There is a danger that projects may be
undertaken not because they are likely to lead to academically interesting
results, but simply because they are susceptible to computerisation.
Scholars feel the need to show they have used the latest technology, even
if electronic methods are not the appropriate one.   But in the words of
Manfred Thaller it is important for humanists to adapt the software to the
scholarship, not the scholarship to the software.
<p no=59 segment_break>
a high proportion of students were delaying job search, mainly because
they planned to continue writing up their theses.   Twelve per cent of
students were not seeking employment at the time of the survey;
<p no=60>
CASE students were rather more likely to be in employment than those with
ordinary studentships;
<p no=61>
just under one in five CASE students in employment were working for the
employing organisation associated with their CASE award;
<p no=62>
the type of department where the student was registered did not always
give a good indication of the research area being covered;
<p no=63>
students registered in an electronics, electrical/electronic or other kind
of engineering department were slightly less likely to be in employment at
the time of the survey than those in other kinds of department;
<p no=64>
research students whose work covered physics, computing or optical fibres
seemed particularly well-placed on the job market;
<p no=65 segment_break>
For simplicity we concentrate on a subset of occam: timing, priority,
vectors, constants, replicators and named processes (procedures) are
omitted.   Our version of occam thus contains only the essential core
needed to write simple programs.   We expect that our work can readily be
extended to versions of the language.   
<p no=66>
For theoretical reasons we will also add a few features to the language:
multiple assignment, output guards in alternatives and divergent ( racing)
process.   IN other respects we will follow the syntax and conventions
introduced in **f, in particular those regarding the parallel operator.   
<p no=67>
(When writing a parallel construct the programmer must declare which
global variables and channels are to be assigned to each component
process.) A finite occam program is one which is WHILE-free.   It may,
however, contain the racing or diverging process **f (equivalent to WHILE
true SKIP).   Much of this paper is concerned with the analysis of finite
programs.   
<p no=68>
This is because the absence of WHILE-loops allows proof by induction.
This restriction does not lose us any power, however, because every occam
program can be identified with the set of its finite syntactic
approximations (a term which is defined precisely in the second section).
<p no=69 segment_break>
the research subject areas where all were in employment were physics,
Computing/physics, computing with non-IT subjects, and in optical fibres;
<p no=70>
research students at work were being employed by 61 different employing
organisations 34 were industrial/commercial organisations and 27 were
educational.   (See Appendix);
<p no=71>
the number of students recruited by any one employer ranged from one to
five;  just over one half of the students in employment were working in
education, as lecturers/teachers or in research;
<p no=72>
of those not employed in education, research students were most numerous
working in the design and manufacture of information systems, software
services and in telecommunications;
<p no=73>
non-CASE students were twice as likely as their CASE peers to be working
in education;
<p no=74>
the 26 students seeking employment were looking in the same kinds of
business areas as those who had been more successful in job search;
<p no=75>
of the 16 students not seeking employment at the time of the survey, 14
planned to carry on writing up their theses, and two had arranged extended
visits overseas.
<p no=76 segment_break>
The first section lists the majority of the laws we require.   We see how
each of the laws arises out of our informal understanding of how occam
constructors work.   We see how algebraic laws allow us to give a precise
and succinct description of each operator.   The laws given are all
congruences in the denotational semantics for occam reported in **f.
<p no=77>
The second section shows how the laws introduced in the first section can
transform every finite program to a form whose only constructs are IF,
ALT, multiple assignment and **f(the diverging process).   Particular
attention is paid to regularising the use of free and bound variables.
We see how this work, together with continuity assumptions, allows us to
prove non-trivial laws additional to those of the first section.
<p no=78>
Even in this restricted form it is possible to write essentially different
programs which are nevertheless semantically equivalent.   The third
section identifies a number of situations where such equivalences can
arise, and develops a normal form for finite programs.   Two normal forms
programs are semantically equivalent if and only if they are syntactically
equivalent in a simple way.   By showing how every finite program can be
transformed to normal form we have thus produced a decision procedure for
the equivalence of arbitrary finite programs.   An infinitary rule based
on syntactic approximation extends this to general programs.
<p no=79>
This proves that our set of algebraic laws (together with the infinitary
rule and substitution) is complete with respect to the given denotational
semantics.   The algebraic laws thus yield an algebraic semantics for
occam that is isomorphic to our chosen denotational semantics.
<p no=80 segment_break>
The major result of the large-scale catalogue use studies was to confirm
the dominance of known-item searching and the role of the catalogue as a
finding tool.  How can we then account for the high occurrence of subject
searches on online catalogues?   There are two possible explanations: the
online environment either created a new demand or stimulated an existing
inherent need.
<p no=81>
Markey's detailed analysis of subject searching in traditional catalogue
use studies reveals that subject searching accounted for between 10% and
62% of searches.   In half of those studies subject searches equalled or
exceeded 40%.   Nevertheless, it would seem that this was not considered
to be a significant proportion.
<p no=82>
The problem of identifying true subject searches is also of particular
interest.   The Yale and Michigan studies found that some searches
initiated as known-item searches could in fact be subject searches or
could develop into subject searches.   The use of a citation as a means of
undertaking a subject search is an established approach.   After all,
citation indexes are based on that principle.   The significance of the
user applying a similar strategy to library catalogues has not been
realized.   
<p no=83>
It is possible that in many studies such hybrid searches have remained
undetected, particularly if users were questioned prior to the catalogue
consultation.   Users looking for information on a particular subject did
not restrict their consultation to the subject portion of the catalogue.
Title searches, for instance, are not  necessarily specific item searches
and could well be subject searches.
<p no=84 segment_break>
But in spite of the multiplication of new basic research tools in the
humanities, it is surprisingly difficult to point, in specific areas, to
solid, uncontroverted gains to scholarship which could not have been
achieved without the new technology.   
<p no=85>
The high hopes which some computer enthusiasts held out that the computer
would revolutionise humanistic study have been proved, over and over
again, to be unrealistic.
<p no=86>
Sometimes the initial claims made were much exaggerated: as the claim made
in New Testament studies that the computer would for the first time reveal
the true lineaments of the Bible that we had ignorantly worshipped.   But
even in areas where there was no hubris in the initial claims, the results
delivered have often been disappointing.
<p no=87>
Between humdrum research and showpiece research, what the humanities
scholarly community is really anxious to see is work which is both (a)
respected as an original scholarly contribution within its own discipline
and (b) could clearly not have been done without a computer.
<p no=88 segment_break>
(1.1) **f This is not an associative law in the usual binary sense of **f
but is analogous in the context of occam's constructors, which can take an
arbitrary finite number of arguments.
<p no=89>
The second law expresses the fact that in the process **f, it is the
first(i.e. lowest index) boolean guard to be true that activates the
corresponding **f Thus **f only runs if **f is true and each of **f... **f
is false.
<p no=90>
(1.2) **f If the boolean guards in **f are pairwise disjoint, then the
order of composition is immaterial.   (This is a symmetry law.)
<p no=91>
(1.3) **f If two booleans guard the same process, they can be amalgamated.
(1.4) **f A false guard is never activated, and so can be discarded.
(1.5) **f If none of the booleans in IF is true, the process behaves like
STOP (i.e. it comes to a complete halt without terminating; a process
sequentially composed with it is not allowed to start).  Thus final
clauses of conditionals which are STOP may freely be added or deleted.
(1.6) **f If one branch of an IF construct is always executed, then the
construct may be replaced by that branch.
<p no=92>
(1.7) **f The final IF law lets us deal with IF constructs which are
nested as processes rather than as conditionals.
<p no=93>
(1.8) **f This law will, of course, be used in combination with
**f, which completes the unnesting.
