BERNARD PARZYSZ

“KNOWING” VS “SEEING”.
PROBLEMS OF THE PLANE REPRESENTATION OF
SPACE GEOMETRY FIGURES

ABSTRACT. There is necessarily a loss of information when moving from a geometrical object
to its drawing. but pupils (even sixth graders) frequently have the illusion that they can, thanks
to a drawing sufficiently sophisticated and close to the object. make a representation of it in
which there is no ambiguity at all.
Similarly, when ‘reading’ a drawing, they tend to regard the properties of the drawing as
properties of the object itself. To illustrate this ‘knowing vs seeing’ conflict, two examples.
studied in three French sixth classes, are given one of ‘coding’, the other of ‘decoding’ a
drawing. They show clearly that it is necessary, at high school level, to make the rules for
drawing space geometrical figures explicit, these rules being not the concern of more or less
‘hazy’ conventions, but of projective geometrical properties.

In France - and perhaps in other countries too - the teaching of space

geometry remains, although the new syllabus endeavours to promote it, the
‘poor relation’ of the teaching of mathematics. It is not given great priority,
it is a part of the syllabus reputed to be difficult among teachers as well as
among pupils and it surely poses some specific problems, bound for not an
inconsiderable part to representation, which can be taken to mean both
mental representation and material representation (drawing) of the figures
which are studied in problems. We shall discuss here this last point, illustrat-
ing it with examples from current research at high school level with ‘scien-
tific’-track sixth graders. This research has connections with those which are
currently being carried out in Grenoble (Bessot, 1987; Osta, 1987), and in
Montpellier (Audibert, 1987; Audibert and Keita, 1987). It is based on the
following assumptions:
(1) There exists a dialectic between the acquisition (or reinforcement) of
knowledge in space geometry, and the mastery of 3D representations.
(2) It is compulsory to pass through a phase of using a 3D representation
(model), even at high school level. We believe it necessary - for various
reasons - for the pupils to learn to do without that kind of representation,
but that can be done only after some time, when the mental images are truly
set up.
(3) There is a necessity of making the rules for drawing space figures
explicit. This type of representation is not the concern of more or less hazy
conventions, but of projective geometrical properties. (Thus, there is a real
opportunity of having the dialectic mentioned above in (1) working . . . .)

Educational Studies in Mathematics 19 (1988) 79-92.

© 1988 by D. Reidel Publishing Company.
80 BERNARD PARZYSZ

Research has been undertaken on:

(1) the more or less implicit principles which underlie decoding (reading)
and coding (producing) plane representations of 3D figures by high schools
pupils.
(2) a didactic engineering which allows these principles to be taken into
account, and to have them evolve towards the conceptual elaboration of a
set of rules consciously mastered by the pupils, giving to their representa-
tions an operating character which they have not got at first, and in return
allowing their knowledge to progress (Colmez, 1984; Bautier et al., 1987).
We shall just mention here the first of these two points, and begin with
fixing the terms used in what follows:
- the FIGURE is the geometrical object which is described by the text
defining it. Cf. Hayward and Sparkes, 1986, for the keyword “figure”: “a
fancy, a creation of the imagination, an idea”. (This is the reason why, in the
present text, we call our illustrations “drawings” instead of using the usual
term of “figures”.)
- this FIGURE is most often REPRESENTED. Cf. Hayward and
Sparkes, 1986, for the keyword “represent”: “to serve as a likeness of (. . .)
to stand for”.
- the representation can be 2D (drawing), if the figure belongs to plane
geometry, 2D or 3D (model) if it belongs to space geometry.
The table below schematizes the relation between the figure and its various
representations. Two levels of representation are distinguished:
- level 1 (close representation): the representation *resembles’ the geomet-
rical figure: same dimension. apart from the move from abstract to concrete.
- level 2 (distant representation): the dimension of the representation is
strictly inferior to that of the figure.

geometry

2D 3D

level 0 figure loss

close repr. level 1 drawing model of

distant repr. level 2 drawing information

There is necessarily a loss of information when moving from a given level

to a higher one, and that loss of information can have various causes.
* level 0 level 1 : Everything cannot be shown on a representation; for
instance. vectors cannot appear. at least in a direct way. Moreover, some
 ‘KNOWING’ VS ‘SEEING’ 81

properties of the representation appear only thanks to the reader’s good will
(restitution of the meaning ). Actually, such a restitution could not take place
without a connivance between the author of the representation (the transmit-
ter) and its reader (the receiver), it being possible only because of a common
geometrical culture. This connivance is concerned in the first place with the
nature of the objects represented, which belong to a restricted number of
types, constituting, so to speak. a reference stock of archetypes (point,
straight line, triangle, circle, plane, pyramid, cylinder . . .), any representa-
tion can then be linked to an -assembly” of such archetypes. This is partic-
ularly evident with computer-created images: for instance. a straight line or
a circle are most exceptionally represented by a drawing which can really
stand visually for what it is supposed to represent, not so much because of
the thickness of the line. as of the ‘thrust faults’ it shows. And, nevertheless,
in a great majority of cases, it is correctly identified.
On the other hand. some figures are not representable, because they are
unlimited (straight line, plane ... ): thus no concrete realization can give an
accurate account of them. The - impossible - representation of such figures
is traditionally replaced by that of a conventional limited part (segment for
a straight line, rectangle for a plane ... ). considered as taking the place of the
whole (geometrical metonymy ). Of course, this can be a source of ambiguity:
for instance, in plane geometry, does a representation. rightly identified as
being that of a segment, really represent a segment. or rather a straight line?
Thus, even at level 1, close to the original. the representation appears
insufficient by nature, like a metaphor in a way, and the necessary interpreta-
tion made by the receiver to give it a meaning might well be improper. Of
course, the weight of tradition helps to lessen such a risk but, on the one
hand it is sometimes insufficient, and on the other hand we shall come back
to that point later on it can be a source of other types of problems.
* level 0 level 2 : This is about drawings representing space figures. The
relationship between a geometrical object and its representation is even
‘hazier’ than in the case of plane geometry, and it often becomes difficult to
guess, from the drawing. some properties of the 3D figure itself. For in-
stance, in Drawing 1A (level 1). representing a square and its diagonals, the
perpendicularity of these can easily be conjectured, on the contrary, in Draw-
ing 1B (level 2). representing a regular pyramid with a square base, the
diagonals of that base and the perpendicular height, it is much less evident
to imagine that the height is perpendicular to the diagonals: one must first
bring to mind a close representation, of the ‘model’ type (and even make it),
if one wants to get to it. unless one is familiar enough with the properties of
3D objects.
82 BERNARD PARZYSZ

Drawing 1.

In conclusion: with close representations, and much more so with distant

representations, it is generally not possible to do without. if not the original
text, at least a caption accompanying the representation, and making up for
the loss of information inherent in any representation: the drawing, the
model cannot substitute itself for the figure.
We shall now take a more particular interest in space geometry, and in
distant representations. Two problems come up then, as commonly happens
whenever communication is at stake: the problem of coding the message, and
that of decoding it.
The problem of decoding has already been evoked above, in a general way,
but here it is complicated by an additional difficulty: that is the danger that
the reader may confuse the drawn 3D figure with a 2D figure having the
same representation. This difficulty is not an imaginary one, and we have met
it, even with sixth graders, for example in an exercise asking. from a drawing
in parallel perspective representing a solid accompanied by a caption identi-
fying it, to tell the nature of one particular side. This side. a square, was
represented by a parallelogram. And. although all the pupils knew that it was
indeed a square, and although the teacher several times laid stress on re-
calling that it was a matter of a quadrilateral in space. and not of its drawing,
however several pupils gave the answer “parallelogram”.
This more or less unconscious identification, tending to consider in every
case the drawing as a close representation of the object. supplies us with an
explanation of the answers given by the pupils (of three ‘scientific’ low-sixth
classes and one sixth class of the Lycée Michelet in Vanves, near Paris) to the
following question, asked among others in a purely perceptual test at the
beginning of the year: “The drawing (cf. Drawing 2 ) represents a plane, P,
 ‘KNOWING’ VS ‘SEEING’ 83

Drawing 2

and three points. A, B, C, in space. For each of these three points tell, from
the drawing, whether it is in the plane P or out of it, or if the drawing does
not allow to decide”. For the 109 pupils involved, the results were as follows:

in P out of P one cannot tell

Point A 81 0 28

Point B 0 78 31

Point C 1 63 45

Of course, it is theoretically impossible to assert whether a given point is

in the plane P or not (and one pupil out of three, on average, is conscious of
that fact), the result is quite clear, as far as perception is concerned: for all
of them, there is an identical interpretation, in the following sense: if the
representation of the point is interior (resp. exterior) to the representation of
the plane, then the point itself is seen as belonging to (resp. out of) the plane.
Let us notice here the possible influence of horizontality (this is another of
our assumptions) in the fact that more pupils think that “we cannot tell” in
the case of point C: the fact that one can imagine more easily extending the
drawing of plane P ‘horizontally’ than extending it ‘obliquely’ highlights
more clearly the ‘contradiction’ between two observations;
- on the one hand, point C is drawn outside the parallelogram;
- on the other hand, if that parallelogram was extended horizontally, the
drawing of point C would be inside the new parallelogram. (N.B. the difference
which can be observed in the distribution between points B and C is statistically
significant at a 95 per cent threshold, but this result should be confirmed).
The problems of coding a 3D geometrical figure into a single drawing have
their origin in the impossibility of giving a close representation of it, and in
the subsequent obligation of ‘falling back’ on a distant representation, in
84 BERNARD PARZYSZ

which there is an additional loss of information (see above). The transmitter

is actually confronted by an insoluble dilemma, due to the fact that what one
knows of a 3D object comes into conflict with what one sees of it. That
unavoidable dilemma has been constant in the history of art, and is summar-
ized as well as can be by Drawing 3 , taken from a fourteenth century Italian
miniature: the anonymous artist has represented the bed as a trapezium (i.e.
what is ‘perceived’) and drawn (from the left) the stripes of the blanket
laying on the bed as parallel bands (i.e. What is ‘known’). Hence the peculiar
look of the right side of the drawing, which concretizes the conflict: the
inferior hand of the blanket is not parallel to the bottom of the bed. Thus,
he who draws a tridimensional object has to find a compromise solution
between the two conflicting, irremediably irreconcilable, poles that are ‘see-
ing’ and ‘knowing’: the ‘primitive’ painters as well as the ‘naive’ ones, for
instance, being rather on the ‘knowing’ side, whereas the ‘classic’ ones, using
the rules of perspective, stand rather on the ‘seeing’ side.
A materialization of the management of this conflict can be found in the
pupils’ productions, as shown in the following example, coming from the
analysis of a test taken in three of the above mentioned classes (two low-
sixths and one sixth):
A close representation of a square-based regular pyramid is used, con-
sisting of a ‘skeleton’ made out of wooden sticks (the edges were 15 cm long,
the height 20 cm, and the square section of the sticks 4 mm). The teacher
describes it, and asks the pupils (after having removed it from their sight)
to make a drawing, with the following purpose: “somebody who knows
nothing about it must be able to recognize that the drawing represents a
square-based regular pyramid”. (N.B. the target cannot be reached without
using additional graphic (conventional) indications: equality of lengths,

Drawing 3
 ‘KNOWING’ VS ‘SEEING’ 85

right angles, etc. Only 18 pupils out of 88 did so, only two of them in a
sufficient way for determining the accurate nature of the represented figure.
This shows that it is not so evident for the pupils, even at such a level, that
the drawing cannot , by itself, substitute for the object.)
We studied the productions of the 88 pupils (30 and 33 in the two low-
sixth classes, and 25 in the sixth class), taking a special interest in the three
following points:
- Position of the image of the base with regard to the edges of the sheet.
- Shape of the image of the base.
- Position of the image of the top with regard to that of the base.

1. Position of the image of the Base with Regard to the Edges of the Sheet

As is usually done. we shall call “horizontal” a straight line parallel to the

upper and lower edges of the sheet, and “vertical” a line parallel to the left
and right edges of the sheet. We have distinguished:

H: at least one of the edges of the quadrilateral representing the base is

“horizontal”.
S: the quadrilateral has a vertical axis of symmetry.
X: other cases.

The results are recorded on the frequency polygons of Drawing 4 (the two
lower-sixth classes are called 1S1 and 1S4, and the sixth class is called TD1).

Drawing 4.
86 BERNARD PARZYSZ

They show that the first case represents at least the three quarters of the total
number, which is not so surprising (it provides the object with a ‘base’), and,
when the quadrilateral is a parallelogram, it corresponds to the position of
the object in “perspective cavalière”, as it is called in French (that is, an
oblique parallel perspective, in which a reference cube has two sides parallel
to the projection plane. this type corresponds in fact to the ‘classical’ repre-
sentation of a cube, drawn backward from one face).

2. Shape of the Image of the Base

We have distinguished:

S. square
D: diamond ( ≠ square)
R. rectangle ( ≠ square)
P: parallelogram ( ≠ diamond, ≠ rectangle)
T: trapezium ( ≠ parallelogram)
X: other quadrilaterals

The frequencies of the different types of production are reported in the

frequency polygons of Drawing 5. A great similarity between the results of
the three classes can be noticed. Besides, the cases which do not pertain to
parallel perspective (i.e. “trapezium” and “other quadrilaterals”) form only
10 per cent of the total number. and in fact are more or less successful
attempts at linear perspective.
The diamond and the parallelogram constitute, for each of them, about 40
per cent of the total number, on average. therefore they account together for
the great majority of the production.
These representations correspond to a ‘side’ view of the object, at variance
with the square and the rectangle. corresponding to a ‘front’ view. This
difference of ‘points of view’ accounts for the small number of pupils having
drawn a square or a rectangular base: in parallel perspective. a front view
glees a lesser depth effect than a side view. Besides, this relative ‘dislike’ is
selective.. it concerns mainly the rectangle (2 per cent of the total number),
which, in comparison to the square (10 per cent), suffers from a major
disadvantage: in fact, the square is a true (close) image of the base of the
pyramid, and this can make up for that, to some extent: the representation of
what is ‘known’ replaces that of what is ‘perceived’.
Similarly., the high frequency of the diamond (which could have been
surprising initially). can he explained by a concern. among some pupils, for
preserving ‘at best’ the properties of the object itself in the drawing. The base
 ‘KNOWING’ VS ‘SEEING’ 87

Drawing 5.

being square, the ideal thing would be to represent it by a square, but such
a picture of a pyramid is much less satisfactory, visually speaking. owing to
a lack of ‘relief’. On the contrary, representing the base by a parallelogram
preserves nothing but the parallelism of the sides. but the perspective effect
is much more convincing. Between these two opposite poles, we find the
diamond which. although preserving the main part of the visual advantage
of the parallelogram. keeps moreover the equality of length of the four sides.
Hence its success (39 per cent).
More precisely, the diamond presents both an advantage and an incon-
venience:
- inconvenience : the ratio of ‘reduction’ in the receding direction equals 1,
and then tends to make the square look too elongated in depth.
- advantage : the conservation of the equality of length for the four sides.
Therefore, the pupils who choose the ‘plain’ parallelogram stand on the
‘seeing’ side (they “present in order to see”), while those who prefer the
diamond stand more on the ‘knowing’ side (the “make known”). The latter
attitude corresponds in fact to the illusion that one can. thanks to a drawing
sufficiently sophisticated and close to the object, and to it alone, make a close
representation of it, without any ambiguity. In the present case, pupils would
have come to an optimal result. and in a more economical way, by the
drawing of a parallelogram. the oblique sides of which being shorter than the
horizontal ones (“present in order to see”), provided an indication of
the equality of length of the sides should have been added, for instance, with
88 BERNARD PARZYSZ

the help of the usual transversal “small bars”, as well as indications of

perpendicularity (“make known”). But that, Obviously, is not a usual and
spontaneous action for our pupils, and needs to be learned. At least, we must
make them become aware of it.

3. Position of the Image of the Top of the Pyramid

We have distinguished:

A: image of the top on a vertical line passing through the intersection of the
diagonals of the quadrilateral (this case corresponds to the ‘classical’
representation).
B: image of the top on the perpendicular bisector of the ‘lower’ side of the
quadrilateral.
C: image of the top on the perpendicular bisector of the ‘upper’ side of the
quadrilateral.
X: other cases.

Remark : The cases A. B and C are not exclusive of one another. In particu-
lar, in a ‘front’ view in parallel perspective, the three conditions are realized
simultaneously (Drawing 6 ). In such a case, we gave the priority to A over B
or C.
The frequency polygons in Drawing 7 show a clear predominance of case
A (about half the total number), which generally corresponds to the classical
representation. But we must also notice the non-negligible frequency of case
B (20 to 25 per cent), which was not expected initially. and which appeared
only at the examination of the productions.
This type of representation corresponds rather (but not inevitably in a
systematical way) to another priority in the construction of the representa-

Drawing 6
 ‘KNOWING’ VS ‘SEEING’ 89

Drawing 7

tion: instead of beginning by drawing the base. one represents first a lateral
side (as an isosceles mangle), and the rest follows. This is to be paralleled
with the classical drawing of a cube in “perspective cavalière” (a drawing
which all pupils of that level are familiar with): you start by drawing a
square. then draw the receding lines and complete.
It is to be noticed that, even in the case of a pyramid, this representation is
quite accurate, from a perspective point of view: it is enough to place the
pyramid so that the reference lateral side is frontal.
The compromise between the representation of ‘what is seen’ and that of
‘what is known’ can be expressed here in the following manner (which is
described usually by ‘mixture of perspectives’; but this expression is not quite
true in the present case):
The ‘mixture of perspectives’ would consist of imagining the pupil as
drawing first the fore side of the object, seen from the front (isosceles tri-
angle) but, becoming aware of the evident lack of relief of his drawing, he
would then transform it by a change of ‘point of view’. However, it
seems more likely to think that, from the beginning, the pupil has in mind
the representation as a whole: if he draws an isosceles triangle, it is not
because he wants to represent the pyramid seen from the front, but rather
because this element seems important to him for the identification of the
represented object: it is a matter of knowledge to be presented. Thus, there is
no interference between two perspectives, but integration of a knowledge into
a (unique) perspective, as when one represents the base of the pyramid by a
diamond.
Besides, let us notice that a representation like that of Drawing 6 preserves
the isosceles shape of two lateral sides, and that of Drawing 1A does so for
90 BERNARD PARZYSZ

the four sides! (In addition to that, these four isosceles triangles are admit-
tedly right-angled. which could not be the case for any pyramid.)
We can see now what proportion of the pupils have represented the pyra-
mid in the ‘classical’ way. This corresponds, let us remind it, to the following
characteristics:
2° P or D: base represented by a parallelogram (possibly a diamond);
3° A: top on the vertical line passing through the centre of the image of the
base.
On the whole there are 33 drawings out of 88 belonging to this type, that
is 37 per cent. the proportion varying from 27 (1S1) to 48 per cent (1S4),
according to the classes.
From that we can conclude that, even if the classical representation be-
longs to these pupils’ competence (they are able to recognize a regular pyra-
mid on a drawing completed with graphic indications), their performance is
not on the same level. for some of their productions hardly evoke a regular
pyramid (Drawing 8 ), even if they are perspectively accurate. For these
pupils. there is actually no conscious and deliberate choice of a perspective,
as well as a position of the object which “presents in order to see” (and
therefore can he used as a ‘support’ for thinking and reasoning), but a use -
more or less important and more or less empirical - of perspective effects,
with a concern for preserving some properties of the object.
Let us go back to the general case. the necessity for managing at best the
‘seeing’ vs ‘knowing’ conflict has led gradually the users of geometry to make
choices which, by reoccurring. have ended in constituting the “tradition”
which we alluded to above. As we said then. there are ‘classical’ drawings,
which can he traced ‘vertically’ from the eighteenth century to nowadays, as
well as ‘horizontally’, from one text-hook to another. That tradition has a
positive side: it allows us to recognize, at first sight, the object in question.
But it has a negative side too, insofar as it might well he restricting: the
pupils. accustomed to drawing a given figure in a given position (for
instance: the cube), lose sight of the ’why’ of that precise representation and
almost don’t think up any others. It is so for the regular tetrahedron,
traditionally represented in “perspective cavalière” with a horizontal side
(Drawing 9A ), and which should sometimes he of greater use if it is inscribed
in a cube (Drawing 9B ).
To conclude, we think that all this shows clearly. the necessity of working,
at least at high school level, on the very principles of the plane representation
of space figures, so its to manage to master them. and not to he a slave of
stereotyped drawings, which have finally lost a great part of their operating
power (cf. Vladimirskii, 1978).
‘KNOWING’ VS ‘SEEING’ 91

Drawing 8.

Drawing 9.
92 BERNARD PARZYSZ

BIBLIOGRAPHY

Adda, J.: 1983, ‘Representations in mathematics’, Actes du Colloque Psychologie et Enseigne-

ment des Mathématiques, Jerusalem.
Adda, J.: forthcoming, Eléments de didactique des mathématiques, Cours Univ. Paris-7.
Audibert, G.: 1985, Une problématique en géométrie dans l’espace, IREM USTL Montpellier.
Audibert, G. and B. Keita 1987, ‘La perspective cavalière et la représentation de l’espace’, Actes
du Colloque CNRS GRECO Didactique et acquisition des connaissances scientifiques, Sèvres.
Bauersfeld, H and W. Zawadowski: 1981, ‘Metaphors and metonymies in the teaching of
mathematics’, occasional paper 11. IDM Universität Bielefeld.
Bautier, T., J. Boudarel, F. Colmez and B. Parzysz: 1987, ‘Représentation plane des figures de
l’espace’, Actes du Colloque CNRS GRECO (see supra).
Bessot, A and M. Eberhard: 1987, ‘Représentations graphiques d’assemblages de cubes et
finalités des situations’, Actes du Colloque CNRS GRECO (see supra).
Bessot, D.: 1983, ‘Problèmes de représentation de l’espace’, Bulletin inter IREM no 23, En-
seignement de la géométrie.
Bonnafé, F.: 1987, ‘Quelques hypothèses et résultats sur l’enseignement de la géométrie de
l’espace à partir de la représentation en perspective cavalière’, Bulletin APMEP (to be
published).
Colmez. F.: 1984, ‘La représentation plane en perspective cavalière des objets de l’espace, un
problème de géométrie, Essai d’ingénierie didactique en classe de Première S’. Actes du
Colloque Inter IREM Géométrie. Journées SMF de Marseille. Ed, IREM de Marseille.
Delorme, A.: 1992, Psychologie de la perception, Ed. Études Vivantes, Montréal.
Hayward, A. L. and J. J. Sparkes: 1984, The Concise English Dictionary, Omega Books.
Osta, I.: 1987, ‘L’outil informatique et l’enseignement de la géométrie dans l’espace, Actes du
Colloque CNRS GRECO (see supra).
Panofsky, E.: 1975, La perspective comme forme symbolique. Ed. de Minuit.
Pérol, C. 1980: ‘Une section du cube’, in Journées Inter IREM de Géométrie d’Orléans. Ed.
IREM de Lille.
Van Sommers, P.: 1984, Drawing and Cognition. Descriptive and Experimental Studies of Graphic
Production Processes, Cambridge University Press.
Vladimirskii, G. A.: 1978, ‘An experimental verification of a method and system of exercises for
developing spatial imagination’, in Soviet Studies in the Psychology of Learning and Teaching
Mathematics 5, The Development of Spatial Abilities, SMSG. Standord Univ. & Chicago, J.
Kilpatrick and I. Wirszup, editors.

IREM & UER de Didactique,

Université Paris 7,
2, Place Jussieu,
75005 Paris,
France