Visual requirements for reading: The importance of a large
field of view in reading with a magnifier


BERRY P.L.M. DEN BRINKER & HUGO BRUGGEMAN

Department of Psychology, Faculty of Human Movement Sciences, Vrije Universiteit, Amsterdam, the
Netherlands & Scientific Institute for Low Vision Use Research (SILVUR), Amsterdam, the Netherlands

Dr. B.P.L.M. den Brinker, Faculty of Human Movement Sciences, Vrije Universiteit Amsterdam, Van der
Boechorststraat 9, 1081BT Amsterdam, the Netherlands, Phone +31204448534, fax +31204448529, e-
mail berry@tbw.vu.nl


Abstract
It is assumed that too low values for optimal field of view in magnifier reading
were obtained in the past by applying the 'Drifting Text' Technique [1,3,4] in
which the subjects had no control of the movements of the magnified image. On
the basis of the view that reading involves alternating sequences of locating and
recognizing textual information [9], it is argued that part of the magnified image
is required for the movement control of the visual display. Higher values are
predicted than the 1-6 characters proposed in the model of Whittaker and Lovie-
Kitchin [1]. 14 Male and female subjects with a macular degeneration ranging in
age from 20 to 82 years of age participated in an experiment to determine the
optimal field of view in CCTV-magnifier reading. Large effects of width and
height are found on reading rate and the data suggest that the optimal values are
even higher than the maximum value of 12 characters that could be technically
realized in the present experiment. Large age effects are found in both reading
rate and smoothness of control of the platform. The data on the movements of
the platform and the eyes are discussed. It is concluded that the elderly subject
applied another strategy to move the platform than the other subjects. In all
subjects large variations are observed in the velocity of transportation of the
platform. It is assumed that these variations signal the flexibility of the motor
control process that is required to adapt the reading process as a whole to
fluctuations in the comprehension process.

Keywords: low vision aids; visual fields; macular degeneration. CCTV magnifier.

Den Brinker. Visual requirements for reading Page 2 of 15
1 Introduction

1.1 The problem of predicting reading rates
By defining the visual requirements for reading tasks, psychophysical models on
reading may improve the efficiency of the selection process of low vision aids.
Whittaker and Lovie- Kitchin [1] recently proposed such a model to predict
reading rate with a diverse set of devices under varying reading conditions. In
their model four visual factors were distinguished: (i) acuity reserve (that is print
size relative to acuity threshold), (ii) contrast reserve (print contrast relative to
contrast threshold), (iii) field of view (number of characters visible), and (iv)
central scotoma size (in case of maculopathy). It is a promising model, allowing
better predictions of reading rate than were possible in the past [2].
However, we have some doubts about the value of one of the parameters
in the model. According to the model an optimal field of view could be obtained
with a 'window' of only 1 to 6 characters. These values were derived mainly from
the studies of Legge and co-workers [3,4] in which reading with a CCTV
magnifier was simulated by having a device move the text in stead of the subject.
It deserves to be mentioned that reading with a fiberscope, a magnifier with a
field of view of 4-5 characters, turned out to be very difficult [5], and later studies
[6,7] found that much wider viewing fields were required for optimal reading
rates. In the study of Lowe and Drasdo [6] the values for optimal window width
were very similar to those found in normal reading [8]. These findings suggest
that higher reading rates with magnifiers can be achieved than predicted by
Whittaker and Lovie- Kitchin, a matter of particular importance for those with
low reading rates who have to decide to use magnifiers for reading or to switch
to another modality (tape or Braille). Therefore an explanation should be sought
for the conflicting results on the role played by the width of viewing field in
magnifier reading.
An explanation for the conflicting results can be found in the
heterogeneous structure of the retina and the consequence of this for the reading
process. The retina with cones in the middle and rods in the other parts of the
retina, affording high central acuity and high peripheral sensitivity, sets limits to
the way the visual system can be used for reading. The lack of full retinal acuity
is compensated for by an ingenuous control system by means of which all parts
of the surrounding world can be focused into the central region [9,10]. For
reading this implies that only a very tiny part of a page can be seen sharply
enough to identify characters. By moving the eyes line segments can be brought
into the foveal area, one by one, making it possible to recognize the characters
and the words and hence comprehend the message. Reading can therefore be
considered as a complex system, in which the control of the eye movements,
word recognition, and language comprehension are integrated subsystems.

Den Brinker. Visual requirements for reading Page 3 of 15
With this knowledge in mind we will introduce new elements for a model
on magnifier reading and offer data to question the remarks by Whittaker and
Lovie-Kitchin [1] on the limited contribution of very wide viewing fields to
improve reading rates. In the remainder of this paper we briefly summarize the
literature on normal reading and reading with magnifiers to discover the
differences and similarities in both ways of reading. On the basis of this analysis
we formulate one single working model for the role of window dimensions in
both types of reading and test a hypothesis derived from that model.

1.2 A working model for reading connected text
During normal reading the eyes make rapid horizontal jumps along the lines of
the text, called saccades. Between the saccades, the eyes are more or less fixated
at the same point and during this period, called the fixation pause, information is
extracted from the visual reading field through character and word recognition.
The saccades are ballistic causing a blurred retinal image from which no
information can be picked up. The length of the jumps is 7-8 characters,
independent (within certain limits) of character size and reading distance. The
fixation pauses vary between 100 and 500 ms with an average of about 250 ms.
Long backward movements, line saccades or return sweeps, occur at the end of a
line when reading has to continue on the next line. Return sweeps are complex
because they are accompanied by a blurred retinal projection of the target at the
moment of their onset. As a result, return sweeps are often misdirected, and
because most of them are too short, they are followed by an extra backward
saccade before reading proceeds [11].
Since extraction of information only occurs during the fixation pause, one
of the major topics in reading research should be the study of the perceptual
span, i.e. the size of the text area from which information is picked up. McConkie
and Rayner developed the so-called Moving Window Technique [12] to study
perceptual span. Subjects had to read from a display while their gaze direction
was monitored. The text was partially manipulated by substituting display
characters by Xs at some distance of the fixation point. Increasing the
undisturbed part of the window symmetrically from one to seven characters did
result in an increase of both saccade length and reading rate. A further increment
of the window did not affect the saccade length but did increase the reading rate
up to a width of 29 characters [12]. The experimenters [13] reran the experiment
with an asymmetrical window. They found that, while reading English, subjects
did not use the area to the left hand side of the fixation point. In subsequent
research it was found that information from the right hand side could be
partially distorted by replacing the characters with a similar shape without
significant effects on reading rate [14].

Den Brinker. Visual requirements for reading Page 4 of 15
The fact that a blurred image of (partially distorted) text still supports
reading performance may be explained by the existence of two parallel
processes: 1) the parafoveal information supports the word-recognition by pre-
processing the visual image, and 2) the parafoveal information supports the eye
movement control system by supplying the targets for the subsequent fixation
point[9]. As will be shown below, similar processes may operate during
magnifier reading.
While reading with an optical or electronic magnifier, the device has to be
moved along the line and, at the same time, kept at a constant distance from the
text to secure a sharp image. Usually the magnifier is moved smoothly to the
right along the line, while the eyes fixate on the same location in the magnified
image and thus move smoothly in the opposite direction. As soon as the
recognition process is completed, the eyes jump rightward to a next fixation
point in the moving image (opto-kinetic nystagmus). This pattern has been found
both in normal and partially sighted subjects [3,4,7,15,16].
Except for the phenomena that hands and eye movements have to be
synchronized to create a stable retinal image, reading with a magnifier is very
similar to normal reading. In both tasks one can speak of a fixation pause to
recognize the characters and a saccade to reach the next point of fixation. In other
words, studying perceptual span in magnifier reading is of the same importance
as it is for normal reading.
Two kind of experimental setups were used to determine the perceptual
span in magnifier reading. Firstly, reading was simulated by the so- called
'Drifting Text Technique' [3]: the magnified text was not moved by the subjects
themselves, but by a device that regulated the smooth transportation of text on
the monitor. Applying this technique, both normal and partially sighted subjects
were found to reach maximum reading rates with a window of only four to five
characters [3,4]. In the second kind of experiment, subjects had to move the text
[6] or the magnifier [7] themselves. Under such conditions the perceptual span
was very similar to the span found in normal reading. A tentative conclusion
may be drawn that wide windows are necessary for the control of the position of
the magnifier, a conclusion that could also be drawn from the experiments with
the fiberscope [5].
We therefore assume that the factor 'dimension' (width and height) of viewing
field is important because it determines the amount of visual information that is
available to select a fixation point for the next word or line. As has become clear
from the research of Rayner and co- workers, the quality of the visual image
outside the 'recognition field' (the area within which recognition takes place)
need not be of a high quality, as long as general properties of the words (their
shape, length) are visible. Furthermore, there are reasons to believe that the
height of the viewing field is also important, especially in the case of the return

Den Brinker. Visual requirements for reading Page 5 of 15
sweep. It is known from the literature that the angle between the end of a line
and the beginning of the next is about 2 degrees [11], while it is also known that
larger errors are made in pointing with the hands [17]. Therefore, it is assumed
that a higher window will help in correcting (with the eyes) misdirection caused
by hand movements [9].
In this study the following predictions will be tested. Within the
limitations of the magnifier system in use, (i) there is a positive relation between
width of the viewing field and reading rate; (ii) there is a positive relation
between the height of the viewing field and the speed with which return sweeps
can be made. To understand the underlying processes, the movements of the
magnifier and the eyes are recorded.


2 Methods

2.1 Subjects
To test the effects of window height and width, reading rates were measured of
subjects with macular degeneration, the category of partially sighted people that
suffer most from low reading rates. An overview of the patients participating in
the experiment is given in table 1. For the analysis the subjects are divided in
three age groups: 20-39, 40-64 and 65 and older. All subjects are experienced
CCTV magnifier users.

Table 1. Subject data.

Den Brinker. Visual requirements for reading Page 6 of 15
2.2 Apparatus
The closed circuit television (CCTV) magnifier used was a TeleSensory
VersiColor XL with a CCD camera and a high contrast black and white monitor
with a diagonal of 46 cm. The text was, with a few exceptions, presented black on
white. The windowing was performed using the functions built in for that
purpose.
To measure the position of the text relative to the camera, the pen of a
graphical tablet (Penpad 300) was attached to the platform, enabling the
recording at a rate of 100 Hz with an accuracy of .05mm.
Positions of subjects’ eyes were monitored with an IRIS system. This system
records eye position by reflection of iris-sclera boundaries by means of infrared
light [18]. The analog output from the IRIS was digitized at a rate of 500 Hz. Both
platform and eye recordings were synchronized.
The head was fixated at an eye-to-monitor distance of 40 cm, using a jaw
bone fixation method. The calibration of the eyes was based on 5 horizontal
markers on the screen. Calibration of the horizontal positions of the eyes was
performed at the beginning of each session.

2.3 Text materials
The text, derived from 'Het Gouden Ei' by Tim Krabbe, was reprinted on A4
format pages with about 18-21 lines. The text was printed in font type 'Helvetica'
(11 point) in lines of 15 cm length, with an average of 80 characters per line, and
an interline distance of 1.2 line height. The text was magnified 13.5 times
resulting in characters (x- size) of 25 mm. width (3.6°) and 28 mm height (4.0°).
This magnification was required to enable all our subjects to read the text.
The smallest size of the window used, corresponded to the size proposed
by Legge and co-workers as optimal for reading drifting text [3,4]. The largest
width was 12 characters, the maximum that could be applied on the monitor. The
three widths were respectively 4, 8 and 12 characters, corresponding with 132
mm (18.3°), 260 mm (33.0°) and 390 mm (44.3°). These widths were combined
with a height of a single line (67.5 mm, 9.6°) or three lines (203 mm, 26.9°),
creating 6 different conditions.

2.4 Procedure
At entrance, all subjects were offered ample opportunity to get acquainted with
the experimental setup, including the CCTV magnifier. Each subject started at
another page in the book and read 7 pages, one page for each of the 6 conditions
and one page for the control condition. In the control condition, with a window
of 12 characters width and 3 lines height, the head was not fixated as it had been
for the other conditions. The order of the conditions was randomized for each
subject. The leader of the experiment placed the text at the beginning of the first

Den Brinker. Visual requirements for reading Page 7 of 15
line and the subject started after the go-signal that marked the digital sampling
of eye and platform movements. Subjects had to read silently and were
interviewed between trials about the story. The leader of the experiment
explained the story of the book up to the page where the subject started. The
subjects had 4 minutes to read a page. Between each trial the leader of the
experiment saved the recorded signals on disk, adapted the window dimension
for the next trial, and explained the lines that could not be read by the subjects.
This inter-trial period lasted an average of 150 seconds.

2.5 Statistical analysis
The first line was not incorporated in the analysis because part of it could be read
before the go- signal. The calculation of the reading rate was based on the
number of words in the lines that were completed during reading. The other
dependent variables, like 'time to read a line' or 'time to return the platform to
the beginning of the next line', are determined on the basis of full lines, excluding
the short lines that mark the end of paragraphs. Differences in mean scores were
tested using Analysis of Variance (ANOVA) with a three factor design (WIDTH,
HEIGHT, and AGE).


3 Results

3.1 Reading rate
Large increases in reading rate are found with an increase in the width
and height of the window and this holds for all age groups (see figure 1 and table
2). The factors WIDTH and HEIGHT are both significant (respectively p < 0.001
and p ≤ 0.001) and are not interacting with one another. Reading rates decrease
with age and this effect is also significant (p ≤ 0.001). Finally there is an
interaction between the factors AGE and WIDTH (p ≤ 0.001) implying that the
effect of an increase in width is greater for the younger groups. However if we
express the advantage of wider windows as a percentage of the mean reading
rate in the narrowest windows the increments are less different for the age
groups, namely 44%, 39% and 34% in the widest windows.
While it is clear now that large dimensions of the view of field facilitate
reading, the subsequent analysis will supply insight in the origin of the beneficial
effects of large windows. The effects of the window dimension are presented on
parameters of the reading process that are often mentioned in the literature: (i)
time required for reading a line and returning the magnifier to the next line, (ii)
velocity of transportation of the text, (iii) smoothness of transportation of the
text, and (iv) the mean length of the saccades.

Den Brinker. Visual requirements for reading Page 8 of 15
3.2 Time to read a line and make a return sweep
The difficulty of finding the beginning of the next line is often mentioned as the
main problem in using magnifiers with small viewing fields [5,7]. Our data also
show that a smaller window takes more time to find the beginning of the next
line. It is found that the return sweeps are faster when the height and the width
of the reading window are increased, effects that are found in all age groups (see
table 3). Both factors WIDTH and HEIGHT have a significant effect on the
duration of the return sweep (p < 0.001; P ≤ 0.005). Furthermore the return sweep
duration increases with age: averaged over all window conditions the durations
are respectively 1.94, 2.52 and 3.53 seconds. However, the factor AGE is
significant at a moderate level (p ≤ 0.036), reflecting relative large variations in
return sweep duration.
Although the effects of window dimensions on the duration of return
sweeps are significant, the size of the effect cannot explain the large differences
found in the reading rates. In other words larger effects are to be found during
the reading phase itself. The mean 'line reading' duration is shown in figure 2 for
the three age groups under the three window widths averaged over window
height. Statistical testing revealed main effects without interactions for the factors
AGE (p ≤ 0.001), WIDTH (p ≤ 0.004) and HEIGHT (p ≤ 0.034).

Table 2. Mean reading rates (WPM) for the subjects in the three age groups reading with
windows with a height of 1 or 3 lines and a width of 4, 8 or 12 characters.

Den Brinker. Visual requirements for reading Page 9 of 15
Table 3. Mean duration (s) of return sweeps (RS) for all age groups reading with
windows with a height of 1 and 3 lines and a width of 4,8 and 12 characters.


To summarize, it is clear that the window dimensions have a larger effect
on the line reading phase than on the duration of the return sweep. For example,
in the oldest age group the mean time required for reading a line decreases from
23.8 in the smallest window condition (WIDTH-4, HEIGHT-1) to 15.4 second in
the largest window condition (WIDTH-12, HEIGHT-3), while the return sweeps
decreased from 5.1 to 2.7 second in the same window conditions.

3.3 The velocity of transportation of text
We now know that larger viewing fields afford higher reading rates and that this
effect is most prominent in the reading phase. The remaining results to be
presented describe kinematic parameters that reflect the underlying (dynamic)
processes in magnifier reading.
From the results presented so far it cannot be derived how the text is
transported during the reading phase. Therefore we analyse the mean velocity in
this section and the variation of velocity in the following section. In the only
study available on this matter, Neve [7] reported that the velocity with which the
loupe was moved forward to read a line and the velocity with which the loupe
was moved backward to the beginning of the next line, were not effected by the
'loupe- width'. This does not contradict his own findings on the reading rates"
since a wider field of view allows less extended movements. However, we
cannot confirm the findings of Neve. The velocity increased with the width and
height of the window and this was the case in all age groups (see also Figure 3).
Statistical testing showed main effects without interactions for the factors AGE (p
≤ 0.001), WIDTH (p < 0.001) and HEIGHT (p ≤ 0.009). In the youngest age group

Den Brinker. Visual requirements for reading Page 10 of 15
the velocity increased from the smallest (WIDTH-4, HEIGHT-1) to the largest
window (WIDTH-12, HEIGHT-3) from 18.4 to 27.2 mm per second. The increase
in the oldest age group in the same window conditions is from 7.5 to 10.0 mm
per second.

Den Brinker. Visual requirements for reading Page 11 of 15
3.4 The smoothness of movement of the platform
The smoothness of transportation of the magnifier (or the text) is an important
'kinematic parameter. It is generally assumed that undisturbed magnifier reading
is characterized by smooth movements of the platform during reading [1,3,4,5,7],
and that most disruptions of the fluency of the movements are related to the
finding of the beginning of the next line [7].
In our study we use the variation of the velocity pattern of the platform as
an indicator of smoothness of the movements. We use COV (coefficient of
variance) to correct for differences in mean velocity. In our case a COV is the
standard deviation of the values of the velocity during a line, expressed as a
percentage of the mean velocity during the same period. COV's are calculated for
each line and averaged for each page.
The results reveal large deviations from the ideal smooth movement
pattern: the mean COV for all window conditions range from about 64% in the
youngest age group to about 187% in the oldest age group (see also figure 4).
Statistical analysis shows a large effect for the factor AGE (p < 0.001) and
a moderate effect for the factor WIDTH (p ≤ 0.039). These large variations in the
velocity are not necessarily a sign of lack of control. Variability may be the result
of a process that requires flexibility. Such flexibility is needed if the reading rate
is disturbed by changes in the comprehension speed. Moreover, since systematic
studies on the movement control are lacking, it is not clear whether or not a
smooth pattern of transporting is optimal under all circumstances [6].

Den Brinker. Visual requirements for reading Page 12 of 15



3.5 Eye movements
Eye movements were only recorded in the window conditions with HEIGHT-I.
The number of saccades per line are calculated and averaged for each page. The
mean number of saccades per line are presented in figure 5. From this figure it is
clear that the number of saccades decreases with the increase of the window
width and increase with age. Statistical analysis reveals a main effect for the
factor AGE (p ≤ 0.008) and a main effect for the factor WIDTH (p ≤ 0.005). The
least number of saccades were made by the subjects in the youngest age group in
the largest window (17.6) and the highest number by the elderly in the smallest
window (68.4), implying that the mean length of the saccades varied between 4.5
and 1.2 characters. However, one should be careful in interpreting these data
because the counting is based on different time samples. When expressed as the
number of saccades per second the differences between age groups and width of
field almost vanish (see table 4).


4 Discussion
From our study it is clear that larger windows are to be preferred above smaller
ones and that the parameter values in the model of Whittaker and Lovie- Kitchin
are too low, at least for people with a macular degeneration. However, the
perceptual span in magnifier reading for these partially sighted people cannot be
derived from this study, since the size of the monitor prevented us from
applying larger windows. Our finding that the relation between reading rate and
window width is almost linear, suggests that reading rates will continue to grow
with wider windows. To find support for this idea, we repeated the experiment
with normal sighted subjects, using a smaller magnification affording a 'larger'
window (24 characters). Under this condition the reading rate still increased with
windows larger than 12 characters, but almost leveled off at the 24 character
window. There is strong evidence, therefore, that the perceptual span in
magnifier reading is of the same order as the perceptual span in normal reading.
We will now interpret the kinematic parameters to understand how the visual
information in the perceptual span is used during magnifier reading.

Den Brinker. Visual requirements for reading Page 13 of 15
Table 4. Mean number of saccades per second (SACC's) for the subjects in the three age
groups reading with windows with a height of 1 and a width of either 4, 8 or 12
characters.


Assuming that reading involves alternating sequences of locating and
recognizing textual information [9], the perceptual span can be divided in two
areas: a recognition field and a location field. Since the frequency of saccades is
more or less stable (in this study about 3.2 hertz), the size of the recognition field
is completely dependent upon the number of characters that can be recognized in
the fixation pause. As a consequence, the maximum extent of the recognition
field is determined by the size and contrast of the characters [19,20]. However, in
our study the mean saccade length, an indicator of the extend of the recognition
field, varied with width and age. In the youngest age group the mean saccade
length ranges between 2.7 characters for the narrowest window and 4.5
characters for the widest window. In the oldest age group these values were
respectively 1.2 and 1.7. These data imply that the recognition field is always
smaller than the available window width. However, it is very unlikely that the
elderly subjects had such small recognition fields since in other studies with
large printed text, elderly people do not behave differently from younger
subjects in this respect [20]. The key to the explanation of these short saccade
lengths are the large variations found in the platform velocity during the reading
phase (see the COV values in figure 4). It is observed that the elderly subjects
more often slowed down the platform during the recognition phase, causing
larger variations in velocity as reflected in higher COV values. In other words,
instead of transporting the platform during the recognition phase, elderly
subjects seem to transport the platform after the recognition has taken place.

Den Brinker. Visual requirements for reading Page 14 of 15
To summarize, we assume that the following mechanisms are at work
during reading with a CCTV magnifier. Firstly, the maximum extend of the
recognition field is determined by the size and contrast of the characters and has
a value that can be calculated on the basis of the contrast sensitivity function for
spatial frequencies [19,20]. Secondly, the recognition field is limited by the size of
the window since space is required for the location field. Thirdly, while most of
the younger subjects prefer to transport the platform during the recognition
phase, many elderly subjects prefer, at least during part of the reading phase, to
transport the platform after the recognition has been completed. Finally, even for
those who are able to transport the platform smoothly, a lot of variation in
velocity is to be observed. We believe that this variation reflects a flexibility that
is required to adapt to fluctuations in the comprehension process.
Since it is shown that a wide field of view is important to attain high
reading rates, manufacturers should reconsider their starting points for
designing low vision aids, especially low vision aids for high magnifications. In a
recent study, [21] it was shown that the present generation of optical magnifiers
were optimally designed for image properties which were inaccessible for their
users. The authors compared the modulation transfer function for spatial
frequencies of various optical magnifiers with the sensitivity for spatial
frequencies of a group of partially sighted people with, as is usually the case,
intact contrast sensitivity in the low frequency domain. They concluded that all
devices tested had wasted transmission performance in the medium and high
frequency range, and recommended designers to sacrifice high resolution for a
greater image diameter. Therefore we do not share the opinion of Whittaker and
Lovie- Kitchin [1] that the physical dimensions of low vision aids do 'NOT
significantly' offer limitations for low vision reading. It was even suggested
previously by us that slight distortions in the periphery of a low vision aid may
be acceptable in cases where wide fields of view are required [22].
To close the discussion, we propose new elements for models on reading
with magnifiers: (i) the eye-hand synchronization, (ii) the dimensions of viewing
field, (iii) the magnifier type, and (iv) the typographical structure of the text to be
read. Eye-hand synchronization is important in two ways. Firstly, when the
magnifier is moved during the recognition phase, imperfections in the eye-hand
coordination cause blurring of the visual image and enforce larger
magnifications. Secondly, when eye-hand coordination is not satisfactory the
reader will switch to another strategy, stopping the magnifier during the
recognition phase. The factors 'width and height' of viewing field are important
since they determine the amount of visual information that is available to select
new fixation points. The next factors, the magnifier type and typographical
structure, set the limits for the previous factors. They determine the quality of the
visual image, the difficulty of control of the movements, and the limits of the

Den Brinker. Visual requirements for reading Page 15 of 15
viewing field. For example in our study the ratio between the duration of return
sweeps and the duration of line reading is much smaller than the one reported
by Neve [7], a difference that is mainly determined by mechanical properties of
the devices used in the studies. Future research should focus on these new
factors to incorporate them in models on reading with magnifiers.

Acknowledgements
H.C. van Capelle, Academisch Centrum Tandheelkunde Amsterdam, Tandtechnisch Laboratorium AK, who
was willing to help us in finding a nice solution for head fixation during eye recording. Dr. G.L. van der
Heijde, Free University, Faculty of Medicine, Department of Medical Physics and Informatics, who enabled
us to record eye movements.

References
1. Whittaker, S.G. and Lovie-Kitchin J. Visual requirements for reading. Optometry and Vision Science 1993; 70: 54- 65.
2. Legge, G.E., Ross J.A., Isenberg L.M., LacMay J.M. Psychophysics of Reading: clinical predictions of low- vision
reading speed. Invest Ophthalmol Vis Sci 1992; 33: 677-687.
3. Legge, G.E., Pelli, D.G., Rubin, C.S. and Schleske M.M. Psychophysics of reading-I, Normal vision. Vision Research
1985a; 25: 239-252.
4. Legge, G.E., Rubin, C.S., Pelli, D.G. and Scleske, M.M. Psychophysics of reading-II, Low vision. Vision Research
1985b; 25: 253-266.
5. Pelli, D.G., Legge. G.E. and Schleske, M.M. Psychophysics of reading III: A fiberscope low vision reading aid.
Investigative Ophthalmology and Visual Science 1985; 26: 751-763.
6. Lowe, J.B. and Dresdo, N. Efficiency in reading with closed-circuit television for low vision, Ophthalmology
Physiological Optics 1990; 10: 225-233.
7. Neve, J.J. On the use of hand-hold magnifiers during reading, Optometry and Vision Science 1989; 66: 440- 449.
8. Rayner, K. and Pollatsek, A. The Psychology of Reading. London: Prentice Hall 1989.
9. Den Brinker. B.P.L.M. and Beek, P.J. Reading with Magnifiers. Ergonomics (in press).
10. Grind, W.A. van de. Limits and mechanisms of vision. In: A.C. Kooijman, P.L. Looijestijn, J.A. Welling and G.J. van
der Wildt (Ed.) Low vision: research and new developments in rehabilitation. IOS Press, Amsterdam. 1994: 28-37.
11. Bouma, H. Visual reading processes and the quality of text displays. In: E. Grandjean and E. Vigliani (Ed.) Ergonomic
aspects of visual display terminals. Taylor and Francis Ltd, London. 1980: 101-114.
12. McConkie, G. W. and Rayner, K. The span of the effective stimulus during a fixation in reading. Perception and
Psychophysics 1975; 17: 578-586.
13. McConkie, G.W. and Rayner, K. Asymmetry of the perceptual span in reading. Bulletin of the Psychonomic Society
1976; 8: 365-368.
14. Rayner, K., Well, A.D., Pollatsek, A. and Bertera, J.H. The availability of useful information to the right of fixation in
reading. Perception and Psychophysics 1982; 31: 537- 550.
15. Bowers, A. and Ackerly, R. Reading characteristics of normal observers using Low Vision Aids. In: A.C. Kooijman,
P.L. Looijestijn, J.A. Welling and GJ. van der Wildt (Ed.) Low vision: research and new developments in
rehabilitation. IOS Press, Amsterdam. 1994: 235-238.
16. Fotinakis, V.P. and Dickinson, C.M. Reading with hand magnifiers. In: A.C. Kooijman, P.L. Looijestijn, J.A. Welling
and GJ. van der Wildt (Ed.) Low vision: research and new developments in rehabilitation. IOS Press, Amsterdam.
1994: 259-268.
17. Graaf, J.B., de Sittig, A.C. and Denier van der Gon, JJ. Misdirections in slow, goal-directed arm movements are not
primarily visually based. Experimental Brain Research 1994; 99: 464-472.
18. Reulen, J.P.H., Marcus, J.I., Koops, D., Vries, F.R. de, Tiesinga, G., Boshuizen. K. and Bos, J. T. The precise recording
of eye moveme: the Iris technique, Part I. Medical Biological Engeneering Computation; 26: 20-26.
19. Legge, G.E., Rubin, G.S. and Luebker, A. Psychophysics of reading-V, The role of contrast in normal vision, Vision
Research 1987; 27: 1165-1177.
20. Bouwhuis, D.G. Reading rate and letter size, IPO Annual progress report 1994; 28: 30-36.
21. Leat, SJ. and Rumney, NJ. Modulation transfer functions of low-vision aids: Comparison with spatial- frequency
requirements in low vision, Applied Optics 1992; 31: 3637-3645.
22. Brinker, B.P.L.M. den. A new approach in designing Low Vision Aids (LVA), inW.L. Zagler, G. Busby and R.R.
Wagner (eds) Lecture Notes in Computer Sciences, 860, Computers for Handicapped Persons, (Heidelberg: