Future Directions for the

Washington, DC Metro Map

A Preliminary Report
Maxwell J Roberts PhD
Department of Psychology
University of Essex
Wivenhoe Park
Colchester
Essex, UK, CO4 3SQ
+44 1206 826923
+44 1206 873797
mjr@essex.ac.uk
www.tubemapcentral.com
Copies of this document can be downloaded from:
www.tubemapcentral.com/WashingtonDC.pdf

Metro
Centre

Gallery
Plaza

LEnfant
Plaza

Future Directions for the Washington, DC Metro Map

Contents
Preface

1. Background: The Washington, DC Metro

2. Prototype designs

3. Proposed methodology

13

4. Conclusions

14

Appendix A: Good design for schematic maps

15

Appendix B: Taxonomy of schematic angle-rules

18

Key bibliography

22

Page 2

Future Directions for the Washington, DC Metro Map

Preface
This document begins with a brief analysis of the Washington,
DC Metro network and its schematic map. Future network
amendments will challenge the current design, and a series of
prototype replacements are introduced and discussed. Each has the
intention of improving on the geometry of the current design, but not
at the expense of excessive topographical distortion. There are many
ways of achieving this, but most graphic designers are somewhat
conservative in their repertoire of solutions.
The current Washington, DC Metro map has a long history and
has become almost synonymous with the network itself. It is
essential, therefore, that any replacement should be carefully
designed and thoroughly researched in order to ensure that it is both
usable and engaging to both Washington, DC residents and visitors.
An iterative, evidence-based methodology is therefore proposed, in
which favored designs are refined in consultation with management
and in conjunction with user studies at all stages.
The appendices at the end of this document give a detailed
discussion of good design in the context of schematic maps, and a
detailed taxonomy of the different types of angle-rules that are
possible, along with their strengths and weaknesses. Examples of
maps worldwide are used to illustrate these.
All official maps are copyright their respective transport
undertakings, and images are used for illustrative purposes only. All
original designs are copyright the author of this document and may
not be reproduced separately from this document without permission.

Page 3

Future Directions for the Washington, DC Metro Map

1. Background: The Washington, DC Metro

The Washington, DC Metro first carried passengers in the
1970s. Conceived of and constructed as a whole, it today comprises
over 80 stations on five lines, with steady expansion taking place [4].
Unlike other networks in the USA, such as the New York Subway
and the Chicago El, the system does not regimentally follow the City
grid structure. Many stations are named after landmarks,
attractions, spaces and districts, rather than the more traditional
practice of naming after intersecting streets. In many ways, the
structure of the network resembles those found in medium-sized
European cities rather than the historic networks of the USA. All of
these factors need to be taken into account when devising a network
map.
1.1 Network map, initial impressions
The map of the Washington, DC Metro dates from the early
days of the network. It is distinctive in many ways but, like almost
all other networks around the world, reality is simplified by removing
most of the details that would otherwise be provided on a standard
topographical streetmap. Other than the lines and stations
themselves, the only remaining details include rivers, parks, county
boundaries, some landmarks, the Washington Beltway and
supplementary station information. Like many other networks, this
is also a schematic map that follows a frequently-used angle-rule
known as octolinear [2, 3]. The complex trajectories taken by the
lines are simplified by converting them into straight lines; horizontal,
vertical, and 45 diagonals (other than the southeastern end of the
Green Line) with tightly radiused corners. The intention is that these
will be easier to follow than if the routes mirrored topographical
reality. The designers undoubtedly hoped that this would result in a
usable, engaging map, conveying the implicit message that journeys
by Metro would be simple, effortless, and rapid.
The most distinctive feature of the map is its thick Metro lines
in relation to the map size and lettering. With future plans
envisaging the possibility of up to three line colors along at least one
corridor, this design style will not be tenable in the future. Less
desirable, station names interrupt the flow of lines in many places.
The adverse effect of this will become more pronounced should line

thickness be thinned in the future. These factors alone point towards

the need for a fundamental redesign, rather than modifications to the
existing version. Station names are also angled in different ways in
many places, and font size and spacing is occasionally reduced so
that certain stations can be accommodated. Typographically, the
overall impression is somewhat chaotic; any implicit message
conveyed by this is almost certainly undesirable from a marketing
point of view.
A more subtle point to note is that the designer of the map has
not taken the opportunity to simplify the trajectories of the lines in
many places. This is an essential procedure in optimization because
substituting the twisting routes of reality with zigzagging routes on a
schematic map does not achieve simplification (see Appendix A).
Page 4

Future Directions for the Washington, DC Metro Map

Instead, the shape of complexity is merely changed. For example, the

Green Line to the southeast of the city changes direction three times
after Waterfront Station. This does not help the user plan a journey,
nor choose the most convenient station. The directional information
is merely noise, defeating the notional advantage of a schematic map.
Journeys by Blue Line appear to be particularly unattractive, with
720 of curvature between LEnfant Plaza and Franconia-Springfield.
Also of note is the twisted route taken by the Green/Yellow Lines
north of Shaw. Two right-angled bends and a 45 one imply a
circuitous journey for the user, who might make the reasonable
assumption that the designer has gone to the effort of implying that
Columbia Heights and Dupont Circle are within easy walking
distance of each other. This information might be useful in the event
of service disruption were it not for the fact that this implication is
incorrect.
Overall, there appears to be scope for improvement in this
diagram. There is ample opportunity for simplifying the structure of
the map and calming its typography, thus enhancing its appeal and
usability. Optional travelers need to be given the impression that the
network is simple, speedy and effortless to use, and one method of
marketing this message is for the map to be likewise.
1.2 Network Topology
Looking at the actual structure of the network, there are a
number of properties that constrain the options for the designer who
is attempting to create the most usable and aesthetic design possible.
It is not a perfectly radial network, and this has implications for how
it is mapped.
The current map is roughly square-shaped with supplementary
information giving it a portrait-format page. However, the network
itself has a strong east-west axis, with the Dulles Airport extension
adding to this. Station names can be difficult to place on lines that
run horizontally. The preferred UK solution is to alternate station
names on either side. Tilted station names are perfectly acceptable in
other countries but should be applied neatly if possible. An
alternative is to slant the lines, but a slope of around 30 to
horizontal is necessary for easy placement of a series of long station
names.

Metro
Centre

Gallery
Plaza

LEnfant
Plaza

Above: Basic Washington, DC Metro central structure and flow

In terms of route trajectories, in the City Centre the lines meet

to form a rough triangle, with the need for clear placement of station
names limiting how this area can be configured. The Green/Yellow
and Red Lines meet, depart, and then meet again further north, and
this reduces the potential for straightening these. To the southwest,
the trajectory taken by the Blue Line constrains the map enormously;
only certain configurations of Yellow and Blue Lines ensure that they
will meet at Pentagon without unreasonably elongated gaps between
stations.
All of these factors should be noted when evaluating the
prototypes shown in Section 2. Particular configurations are chosen
with good reason, and alternatives may be impractical.
1.3 Topographical considerations
Many users complain about topographical distortions on
schematic maps, although the difficulties that these cause are
probably overstated. Overall, the designer should distort topography
where this will improve the flow of the lines, but avoid doing so
Page 5

Future Directions for the Washington, DC Metro Map

excessively. Whether a user is seeking to identify a start or

destination station, or alternatives in the event of service disruption,
it should always be remembered that a network schematic is a poor
substitute for a detailed streetmap, and direction signage and maps
of local neighborhoods should ideally be provided at each station.
The Washington, DC Metro has a number of properties that
mean that stringent topographical accuracy is less important than
for other USA urban rail networks.
All stations are uniquely named. In New York and Chicago, the
naming of stations by intersecting street often results in
duplicated names. A topographical map is helpful in resolving
such ambiguities, but the Washington, DC Metro has none.
Almost all likely transfers are catered for within stations. For
some networks, e.g. the London Underground, certain journeys
may be completed more quickly by between-station transfers;
leaving a train at one station and transferring to another
station a few hundred yards distant. Planned as a whole, the
Washington, DC Metro caters for most possible journeys via its
key transfer stations, and so there is no need for the map to
indicate nearby stations that could be used for transfer
purposes. The only possible exception to this is that a journey
from, say, Shady Grove to Clarendon might be speeded by
transferring from Farragut North to Farragut West, but the
time advantage is likely to be small, and this example is
probably a rarity.
The most appropriate station is clear for most suburban
districts. Many suburbs in London and New York are served by
multiple lines at different stations, so that the user may need
to be made aware that there are nearby alternatives. In
Washington, DC, few suburbs have competing services, and so
the map need not highlight these.
Overall, considering topographical fidelity, the designer of the
Washington, DC map has more leeway than most USA networks.
Gross distortions are to be avoided because these will be resisted by
users, but there is little in the way of travel hints and suggestions
that need to be communicated via spatial configurations on the map.
Page 6

Future Directions for the Washington, DC Metro Map

2. Prototype designs
Several prototypes have been developed and these are
discussed in this section. An explanation of the different angle-rules
that schematic maps may follow is given in Appendix B, where the
nomenclature adopted is also explained more fully. For the
prototypes: size, shape and font are matched to the current design.
They have also been matched for basic supplementary information
(parks, rivers and landmarks) but more detailed information, such as
county boundaries, has not been included. This will be added at
future stages of the design process. Design aspects such as line
thickness, colors, and (uncontroversial) station and transfer symbols
have also been matched so that the maps can be easily compared and
evaluated. None of these prototypes should be regarded as a
definitive final version. Ideally, a systematic, evidence-based
methodology should be adopted, as described in Section 3.
Section 3 explains that the aim of the early design stages (1 to
3) is to create alternative maps and then identify from users the most
favored of these. If too much supplementary information is included
at this stage, it will be difficult for users to evaluate the basic
designs. Subsequently, the most favored alternatives will be refined
(Stages 4 and 5) and provision will be made on these to include
further supplementary information. However, this will only be added
at later stages (6 onwards) once the preferred network configuration
is finalized. From here, alternatives for station and transfer symbols,
etc., can also be investigated. Hence, in all studies at all stages, users
can give their full attention to the key questions under investigation.
For all the prototypes, station names are unabbreviated and
none interrupt lines. The Dulles Airport extension is included in full,
and the possibility of a separate designation for the proposed service
from Franconia-Springfield to Greenbelt is adopted. The prototypes
are ordered step by step in increasing complexity of the angle-rules
adopted.

Page 7

Future Directions for the Washington, DC Metro Map

2.1 Tetralinear Map: Two angles

(horizontal/vertical) rotated by 30
Tetralinear designs are too severe for
general use, all too often resulting in
many corners, considerable
topographical distortion, or both.
Occasionally, a rotated design can be
visually striking. This is the case for
the Washington, DC Metro, where a
very distinctive creation is possible.
This is the only prototype of the five
where none of the station names is
tilted.
Advantages
Clean, visually striking, good
typography. Good trajectories for Red
and Green Lines.
Disadvantages
Topographical distortion, users may
resist such an abstract design.
Corners are severe (90 turns).

Page 8

Future Directions for the Washington, DC Metro Map

2.2 Hexalinear Map: Three angles

(horizontal, 60 left to right, and 60
right to left)
Hexalinear maps can look very
dynamic when matched to the right
network. For Washington, DC, the
steep angle results in some
difficulties in meeting the Blue and
Yellow Lines, hence the hump at
Rosslyn. On the other hand, the lines
fan into the suburbs in a satisfactory
way, and the inherent compactness of
60 angles enables the Dulles airport
extension to be shown cleanly.
Advantages
Lines fan cleanly into SE suburbs,
Yellow/Purple/Green and Red Lines
have excellent trajectories. Design is
reasonably compact horizontally.
Hexalinear angles unlikely to
surprise or be resisted by users.
Disadvantages
Angles do not quite fit the W/SW part
of the network.

Page 9

Future Directions for the Washington, DC Metro Map

2.3 Octolinear Map: Four angles

(horizontal, vertical, and 45
diagonals x2)
The traditional angle-rule for a
schematic map fits the Washington,
DC Metro reasonably well, and
considerable improvements are
possible over the current official map
in terms of geometry, particularly for
the east-west lines. However, the Red
Line fares less well, and the lack of
compactness forces a bend in the
Dulles Airport extension.
Advantages
Standard design technique, therefore
likely to be widely accepted by users.
East-west and Blue Lines have good
trajectories. Low levels of
topographical distortion.
Disadvantages
Extra corners required elsewhere,
little overall advantage over the
hexalinear map in this respect.
Dulles Airport extension suffers.

Page 10

Future Directions for the Washington, DC Metro Map

2.4 Higher-Order Linear Map: Based

upon 22.5 increments, technically
Hexadecalinear
Higher-order linear maps are difficult
to design and should only be
attempted if there is a clear objective
that may be achieved in doing so.
Even so, results can be disorientating
for users. In this case, the objective
was to force a perfectly straight
route from Dulles Airport to the city
center. The minimum tilt to permit
this is around 22.5 (shallower angles
result in difficulties placing central
stations). Overall, the city is now
slanted, which some users might
object to, although the lines
themselves have the best trajectories
of the four prototypes so far.
Advantages
Fewest corners, new extension is on a
straight line with clear publicity
value.
Disadvantages
Unusual design might meet with
resistance. Largest number of tilted
station names of all prototypes,
although these are arranged neatly.

Page 11

Future Directions for the Washington, DC Metro Map

2.5 Curvilinear Map: No straight

lines permitted
Curvilinear maps enable harsh
corners to be smoothed away,
replaced by gentle curves. Some users
initially find this style unsettling,
although many see an immediate
appeal. In general, this type of design
has the greatest advantage over
traditional schematics for very
complicated networks, but the
Washington, DC Metro version
demonstrates its potential well.
Advantages
Novel design. Corners smoothed away
without excessive topographical
distortion. Appeals to those users who
find strict geometry harsh and
intimidating.
Disadvantages
Many users (around 4050% in the
UK) prefer strict geometry, although
this proportion will be greatest where
the tradition of octolinear schematic
designs is greatest, such as London.

Page 12

Future Directions for the Washington, DC Metro Map

3. Proposed methodology
As shown in Section 2, there are many different angle-rules
that a designer can use to create a network schematic, each with
advantages and disadvantages that vary by network. It is therefore
important that the designer takes a systematic approach. Different
angle-rules must be explored and used to devise prototypes. In the
process of this, the designer can identify any points of difficulty,
along with particular angle-rules that assist or exacerbate these.
User studies are also important, and cost-effective to conduct
in relation to the entire design process. A number of these have been
conducted at the University of Essex and London South Bank
University in the UK (e.g., [3]). One major finding is that subjective
ratings of designs (e.g., an estimation of how easily lines can be
followed) have no relationship with objective usability measures
(such as speed of planning a journey between two stations).
Essentially, map engagement is entirely separate from map usability,
so that an individual may dislike a map that he/she finds easy to use,
or like a map that he/she finds difficult to use. Of course, a skilled
designer will aim for an engaging, usable design, but it is important
that user studies evaluate both aspects of behavior.
The overall plan is to design a map in stages, weed out the
least liked prototypes, optimize the basic structure of the remainder,
establish the most favored design and then determine fine details
such as station/transfer symbols and supplementary information.
The sequence of stages would be as follows.
3.1 Preliminary stages
The designer should survey the network if possible, and consult
with users and management concerning the strengths and
weaknesses of existing designs. Requirements such as font/font size,
dimensions, colors, and priorities such as keeping certain lines
straight but not others, should be established early if possible.
3.2 Rapid prototyping
A range of prototypes following different angle-rules should be
produced. These will be intended to be well-optimized designs, but
subject to future refinement. At least four should be created:
traditional octolinear, curvilinear, and two others based upon those
angle-rules that match the particular network well.

3.3 Preference studies

In order to narrow down the options, users will be asked to
rank-order designs for preference, and describe the perceived
strengths and weaknesses for each. To assess the wider impact of
different designs, ideally at least three groups of users should be
consulted: (a) City experts, (b) Transit experts who are not familiar
with the particular city, and (c) transit/city novices.
3.4 Refinement
On the basis of user ratings, no more than the two most
favored designs should be selected for further optimization. Different
options can be explored, and final network configurations will be
aimed for. If any prerequisites concerning supplementary
information, etc., are known, the designs should be made ready to
accommodate these, although they would not be displayed on the test
designs until 3.6.
3.5 User testing
Usability studies should be conducted on different groups to
ensure that the refined designs are usable and have no pitfalls
(modifications are possible in the light of this). Assuming that they
are evenly matched for usability, further preference studies will
determine the most favored design.
3.6 Embellishment
Once the basic network configuration is finalized for the most
favored option, further choices need to be made concerning station/
transfer symbols and supplementary information. If necessary, these
can be evaluated via further preference studies (which are
straightforward to conduct).
3.7 Adoption of final design
Consultation with users and management at all stages, along
with a systematic evidence-based approach, should ensure the
widespread acceptance of the new design by the general public and
the media.
Page 13

Future Directions for the Washington, DC Metro Map

4. Conclusions
A series of prototypes has been created, intended to optimize
the line trajectories on the Washington, DC Metro schematic, and
therefore usability and appearance. Each is intended to be possible
candidate for the basis of a future official map, and to demonstrate
the level of improvement that is possible. Ideally, the next step is to
evaluate user preferences. An evidence-based, consultative, iterative
methodology is proposed. By developing the favored designs in this
way, continuing to engage users throughout the design process via
journey planning, preference, and subjective evaluation studies, a
distinctive, attractive usable design will result. This will gain
widespread acceptance soon after introduction.

Page 14

!3;*%--,
0#'1

4%$2-%6,0#'1

*+'!",
4%$2-%6

*%#&/%*
>,1(*.&23'6

4%&!
"#$0&!%#/

/+--(&,
"(--

4%$2-%6

4(--%&/%*,
.'%%*

APPENDIX A: Good design for schematic maps

!+

:3%%*&,
0#'1

#-0%'!+*
"#'-%&/%*

"(."23'6,>,
(&-(*.!+*

;#'$
5#*+*23'6,>,
Future
Directions for the 6+'1,'+#/
Washington, DC
Metro Map
%&&%?,'+#/
5#$/%*,
!+4*

1(-23'*,
0#'1

!+
&!%0*%6,.'%%*
$(-%,%*/
2+4,'+#/
2'+$-%6
4%&!,"#$
0-#(&!+4
30!+*,0#'1
%#&!,"#$
2#'1(*.
30*%6
2%5+*!'%%
"%#!"4#6
/#.%*"#$
"+'*5"3'5"
30$(*&!%'
#*/,&+3!"%*/

5"#-1,

&4(&&,
5+!!#.%

&!+*%2'(/.%,0#'1

0#$)6)4-#)
1$&51&)-

$#'-2+'+3.",
'+#/
.'%#!,
0+'!-#*/
2#1%', &!'%%!
&!9,<+"*&,
&!'%%!
4++/

%3&!+*

1(*.&,5'+&&
&!9,0#*5'#&

#*.%-

+-/,
&!'%%!
%3&!+*,
;#''(*./+*
&:3#'% '3&&%--,
4"(!%5"#0%4#''%*,
$#'6-%2+*%
&:9
#-/%'&.#!%
4#'4(51,
4##*4&%5-'
4(--%&/%*,
4%&!2+3'*%,
&!9
$++'.#!% !"#$%&'%(#)*#)%$(+,
*+'!"
#)%*3%
<3*5!(+*
5"#*5%'6
2'(!(&"
0#'1
2(&"+0&,
'%.%*!&,
#5!+*
-()%'0++-#*%
$3&%3$
'+#/
0#'1
-#--&)./0
0+&!,+;;(5%
&!'%%!
-#/2'+1%,
'+6#-,+#1
+?;+'/ 1#2$-%$*'
&!9,$#'6&
.'+)%
&*49/$&% 5('53&
"+-2+'*
0#//(*.!+*
$#*&(+*,
$#/*
2#*1
#-/.#!%,%#&!
4%&!
#-/465"
-#!($%','+#/
2#6&4#!%'
7$/&*%5-'
"+3&%
#5!+*
-%(5%&!%',
#-/.#!%
2+*/
*+!!(*.,
1#3&)-%
&:3#'%
$#'1,-#*%
$#'2-% &!9
3?2'(/.% "(--,.#!%
4/$*&)
2-#51;'(#'&
%#&!,
4++/,
-#*5#&!%', #'5"
5#**+*
'+#/
$+*3$%*!
#5!+*
!%$0-%
:3%%*&,
-#*%
%#-(*.
.#!%
%1%0#1()"!(
5-$/)*
-+)#1!&*@D$%)'1$
0(55#/(--6
&!'%%!
%#-(*.,-+)#1!&
12"$*
'+#/
2'+#/4#6
5"#'(*.,5'+&&
5+$$+*
/+)%',&!9
7'&(D77"!*
-+*/+*,2'(/.%
-+)#1!&*95'35*()!""):
"+--#*/,
&+3!",
;<==
"$#'",&*-"()
%!1?,".*3!""$
!'%07+$(-1!)5
!'#;#-.#'
4#!%'-++
&"%0"%'/&,23&"
0#'1
%#-(*.
"6/%,0#'1,5+'*%'
#5!+*,
&:3#'%
/D(5".*+$&*1?5".
!+4*
2+'+3."
.+-/"#41,
"(.",&!'%%!
711!*2+!0*
(1D)53+)"
4%&!$(*&!%'
&+3!"
'+#/
%+!2"$&"!(*2+!0
1%*&(*.!+*
6*(+$&.*,1&3"
#//(&+*
1*(."!&2'(/.%
#5!+*
()+$71!"
'+#/
%-%0"#*!,
.-+35%&!%'
!3'*"#$
+!$1(*3!1E"
'#)%*&5+3'!,
2'+$0!+*,
5+)%5*"$&*#1!* 2'$$"!
$1!)5-11&
*+'!";(%-/&
>,5#&!-%
'+#/
&!9,<#$%&
-#$2%!"
.'%%*
%#'-&
0#'1
'+#/
2#'+*&
*+'!"
"&3-+!"
)(5!+'(# 0#'1
5+3'!
%+$1$(*2+!0
5"+&()1$"*,+$"
/1D$&(*3!""$
5+3'!
2'$$"!
5"(&4(51,0#'1
1%**(*.!+*
/D!$)*1+0*9-+),'$3:
!"..052925'"*(%&03'
&+3!"
&!#$;+'/,
!+
0'$3(/D!.
&-+#*%
,'3&03
-11&*3!""$
1%*&(*.!+*%1,'$&+,"
2'++1
2+&!+*,$#*+'
&:3#'%
.3**%'&23'6
2!"()1$*
5+!!1-*
$1!)5*
5'353+)"
4%&!
+&!%'-%6
4%&!,
$"+(&"$
!1+&
1$G)5"G5',, "#$$%'&$(!"
5+!!1+)#5"$&1$*%"$)!+,
1%*&(*.!+*
"+3*&-+4,%#&!
2'+$0!+*
)D!$2'0"*,+$"
&1,,'(*5',,
"+3*&-+4,5%*!'#!"#$%&
'()%'/!"$)
30.&!2
$1!)5-'%0*
-"7/,".*
"+3*&-+4,4%&!!D'(,'2* 1%4,.#'/%*&
!+.$"!(*
&!+514%-)D#$",,*
5'#)*'1
2+!0
-"()*
2+!0
4#-"#$,.'%%*
31,&"!(*
7+$1!*51D("
-',,"(&"$*3!""$
D?/!'&3" '%0"$5+7
7+$1!
,+$"
2+!0
5+!!1-"7/,".*#1!* (D&/D!.
3!""$
5-#0"#$,*+'!"
()1$"/!'&3"*2+!0 0#'&+*&,.'%%*
0',/D!$*6*/!1$&"(/D!.
!"#$%&'"(
!D'(,'2
"+()%1)"
5',,'$3&1$
(1D)5*5+!!10"$)'(5* 5-#0"#$,5+$$+* #'$(/D!.*2+!0
5+!,"(&"$
)1-$
03!*%6,2'(/.% -"()*5+72()"+& /",('>"*
2+!0
-',,"(&"$*@D$%)'1$
5-#0"#$,&+3!" +!("$+,*
(D&/D!.*5',,
&!+.)1$*2+!0
'(5"$+*/
&3/23'6,"(-&!","#''+4
'#6*%'&,-#*%
%#&!5+!%
'3(&-(0,$#*+'
'3(&-(0
(51%*"#$
"(--(*./+*
#*/,3?2'(/.%
*+'!",%#-(*.

Since the 1930s, transport undertakings around the world have

been displaying their networks by using simplified diagrammatic
depictions. Such schematic maps are intended to enable people to
plan journeys without having to contend with irrelevant distractions.
They are also important publicity devices, encouraging system usage.
To attract people for whom travel is optional, it is important that
schematic maps communicate the overall impression that a journey
will be simple, effortless and rapid. The difference between a good
and a bad design will contribute a great deal to first impressions
formed by the potential user.
Simplification

1(-23'*,>,
2'+*/%&23'6

;(*5"-%6,
'+#/

/'#6!+*,0#'1

5#-%/+*(#*,
'+#/

1%*!(&",
!+4*

2%-&(=%,0#'1

;+',&3/23'6

52*82$+%
-#9)

"#$0&!%#/

"+--+4#6,
'+#/

1%*&#-,
.'%%*

$#(/#,
)#-%

0#'1,
'+6#-

Above: Topographical map of the London Underground in the style issued

before 1933. Below: The schematic map of the London Underground
designed by Henry Beck (1933) simplifies the trajectories taken by the
lines, making the routes easier
to follow
and the underlying structure of the
#'$%5,".*!1+&
%5+,0*
0"$(+,*3!""$
#+!7
%#&!,03!*%6
2#-"#$
network easier
to
identify
and
learn.
-909!929
(-'((*%1))+3"
5'35/D!.*6*'(,'$3)1$
%+7&"$*
51,,1-+.*
FD""$(*2+!0

A precise definition of a schematic map is difficult to identify,

/1
but a browse through a selection of network maps [1] will reveal
/!
'%;%'%*5%(D&/D!.*)1-$
)1-$
!1+&
!'(*(!6,'+#/,
&+3!";(%-/&
/(&!'(5!,,-(*%
$%!'+0+-(!#*,,-(*%
2,
0',/D!$*2+!0
7!++!(*.,2%58
7+!,/1!1*!1+&
some clear patterns across the world. Many undertakings
use
a
+,2"!)1$
D
2#1%'-++,,-(*%
$%!'+0+-(!#*,,-(*%
71!$'$3)1$* %+,"&1$'+$*
"+
%+$1$/D!.*6*"(("?*!1+&
7+'&+*E+,"
!1+&
%!"(%"$)
0(55#/(--6,,-(*%
!++!(*.,2'+#/4#6
/+
4($2-%/+*,0#'1
0'$3(*%!1((*
2+!0*!1.+,
topographical map. This might simplify reality to%/.4#'%A,,"(.".#!%
some degree, for
5%*!'#-,,-+*/+*,,-(*%
D
-+!-'%0*+E"$D"
7+!.,"/1$"
()8*2+$%!+(
,+)'7"!* -"()/1D!$"*
"D()1$
/"
5+--(%'&,4++/
>,,$+'/%*,,-(*%
!1+&
2+!0
1,&*()!"")
5
"+,'$3*
$1!)5*"+,'$3
example by omitting all but the most important landmarks,
but
the
(*!%'5"#*.%,,&!#!(+*&
&
+$3",
&+3!",4($2-%/+*
/!1+&-+.
"D()1$*
51
,+&/!10"*
/+0"!*
7$%'!+*8
/'(512(*!1+&
"+()*
-"()*
(FD+!"
D
4($2-%/+*
3!1E"
711!3+)"
()!"")
+%)1$
-+!!"$*
locations of stations, and sometimes even the trajectories
of the lines +%)1$
/+.(-+)"! 2+&&'$3)1$ $+'/%*
6
2!+"&*
(&&3%/,,26,,-+*/+*,,0#&&%*.%',,!'#*&0+'!,,2+#'/,,@@A,,2'+#/4#6A,,&949B
()!"")
!D((",,*
!"3"$)(*$#?4%--9,<9,'+2%'!&A,BCDEFDFEEGA,2#&%/,30+*,/%&(.*&,26,;9"9,&!(*.%$+'%
()!"")
(FD+!"
311&3"*
$1!)5* -11&*
2+!0
,'E"!211,*()!"")
$1))'$3*
,+$%+()"!*
%5+$%"!.*
themselves will mimic reality as much as is possible on a scale
21()*
/1$&*
()!"")
+%)1$ ,+$"
D?/!'&3"* 5',,*3+)"
3+)"
,+$" 1##'%" /+$0
()!"")
"+,'$3* (5"25"!&(*
!1+&
(51!"&')%
drawing. The problem with such maps is that the complex
twisting of%1771$ /D(5
/!')'(5*
51,,+$&*
FD""$(*
7+!/,"* 1?#1!&*
+,&-.%5
7D("D7
%'!%D(
(1D)5*"+,'$3
2+!0
!1+&
+!%5
+,&3+)"
%1E"$)*7+$('1$*
()"2
lines adds information that is irrelevant to journey planning
(what
is
3+!&"$*
()8*7+!.(
3!
51D("
31,&5+-0*
$1!)5#'",&(
!1+&
2'%%+&',,.
/1()1$*7+$1!
7+!0*
&1E"!*()!"")
+,&3+)"*
,"'%"()"!*
+&&'(1$*
the best way to get from Station A to Station B?) and 1()"!,".
can mask the +%)1$*)1-$
5'35*()!"")*
%&."3(
,+$"
"+()
(FD+!"
!1+&
0"$('$3)1$ 5.&"*2+!0*%1!$"!
)"72,"
(1D)5*+%)1$
%5+!'$3*
(5+&-",,
0$'35)(/!'&3"
51D$(,1-*"+() routes
)!+#+,3+!*
overall structure of the network, so that the underlying
of the ()+7#1!&* 5+77"!(7')5
%5'(-'%0*
3,1D%"()"!* /!172)1$*
%!1((
(FD+!"
"+!,(*
2+!0
51D$(,1-*%"$)!+,
,1$&1$*/!'&3"
!1+&*
/+!1$(*%1D!)
-+22'$3
-"()7'$()"!
!1+&
%1D!)
/!110
-+)"!,11
E'%)1!'+
51D$(,1-*-"()
lines are harder to identify and learn.
,+7/")5*
3D$$"!(/D!.
:!/4!)6.12!/**;

7.'%#!,*+'!"%'*,>,5(!6,&%5!(+*8

%&7280!3%2500(

.01"*2
0",

'(45".'2
.0"(

4.'"&2
$0.&*"3(2
%&.''&

-"../34(03

"*('.%4"&'

!0*)0.3

&0&&'3!"#2
+06.&2.0"(

#036#'3&

)*"+,-./".% +"33032
%&.''&

)D!$5+7* !+E"$(%1D!)*
2+!0
3!""$

-"()*
0"$('$3)1$

(1D)5*
0"$('$3)1$
-"()*
/!172)1$

(,1+$"*
(FD+!"

$1!)5

()8*@+7"(*
2+!0

5!/&'+!"

/1!1D35
","25+$)*
6*%+(),"

!1)5"!5')5"

(D!!".*&1%0(
The alternative to a topographical map, a schematic map 0"-*3+!&"$(
seeks
-+,5+7*3!""$
0"$$'$3)1$
to simplify reality further by converting the complex trajectories
of
!'%571$&
Design rules versus
design objectives
2+!(1$(*3!""$
1E+,
$"-*%!1((
routes into simple straight lines. Similar to the design used by the
()1%0-",,
3+)"
2D)$".*/!'&3"
%,+25+7*$1!)5
London Underground for over 75 years, corners are tightly radiused,
A schematic map can assist the user if the
complexities
of
%,+25+7*%1771$
"+()*2D)$".
and a limited number of angles is permitted, almost always
!"#"!"$%"reality are genuinely
%,+25+7*(1D)5
simplified. This is important
because there is a
&'()!'%)**!+',-+.
7")!121,')+$**!,.8
/+,5+7
/+0"!,11**,'$"
7")!121,')+$**!,.8
horizontal/vertical lines along with diagonals at 45 (this particular
(1D)5#'",&(
misconception
amongst
graphic
designers
that
octolinear
by itself
2'%%+&',,.**,'$"
)!'$').*!1+&*
UndergrounD
"+()**,1$&1$**!+',-+.
-'7/,"&1$*2+!0
combination of angles, or angle-rule is known as octolinear,"&3-+!"4**5'353+)"
see
)11)'$3*/!1+&-+.
'$)"!%5+$3"**()+)'1$(
6**71!&"$**,'$"
constitutes
some
sort
of
cartographical
gold
standard,
such
that maps
%"$)!+,**,1$&1$**!,.8
D$&"!*%1$()!D%)'1$
%1,,'"!(*-11&
Appendix B). There the similarity ends, and there is a bewildering
-'7/,"&1$
designed using this
angle-rule will automatically
be easy-to-use, and
(1D)5*-'7/,"&1$*
variety of techniques for grouping services and showing stations,
that maps that fail to use this will 71!&"$
inevitably have a usability
58*%8*/"%04*;<==8*&'3')'>"&*/.*7+?-",,*@8*!1/"!)(4*;AB=BCAA<
transfers, and other information [1]. However,
the basic octolinear
disadvantage. This belief is easy to refute both theoretically and
straight-line technique for showing the routes of the network
empirically. If we consider that the goal of the designer is simplicity,
remains dominant.
then bad implementation will turn the meandering curves of reality
93!"+)*$1!)5"!$*6*%').*("%)'1$:

9)11)'$3*/"%:

97"!)1$:

into zigzagging corners on a map. Reality has not been simplified,

Page 15

!A<D>
*NCEGCM
>0%()*;$#B8$;$#B8$-*,$/(%"8

EIB%'14$
>0%()08(

R51B5%-*!'88
3]O>CQB9

V%$$/D0%1

Z$/('&)
2A>G9EQ9<

!'88
E@@$%*!0880J"- $CGG9EQ9<'W>99< ]'/&B5%-*="%:

,>D<Q9EON>:'
T5D/$88*="%:
PA>K

;"8()"#&(0J
2AIME=9AQ
^5$$/&*K0"1
2AIME=9AQ'
WDEM9G'
29A=@
+AK

UC<?@G9:'*DAQ'
J'U>DB<AG

8CGON><

T0J/
;$&(89<EAG'*CE9
,>D<Q9EON>:
Z$/('&)*T0J/

C%&$/"8

3MM9>

-NV<9GG'P

$9E='2AIME=9AQ

89<=CE@'
]'/7)8$- ,)"8:*]"%#
,9GEC`9'PA>K
89<=CE@'-DS<
-DS<'
!0880J"$CGG9EQ9<'(N<?=CD<
K0"1*+
$9E=
K0"1
FAIQ9<'
UC<?@G9:'*DAQ
]%04/"8
89<EAG'W>99<
*DAQ
F@AGK'UA>I
,"#1$/*K0"1
6"8&(0/
,"8$10/'"/*K0"1
,"#1$/*T0J/
8CGON><'
FAIQ9<'
,"/0/B5%Z'/4&8"/1
FAG9QD
+*9"%/&B5%\N99<ZE'PA>K 2CB@'*DAQ
.SCEE'FD==AB9
-DS<
]'/7)8$-*K0"1

Z'8B5%/
Z$/&"8*K'&$ .NQON>:'2CGG
;$&(*!"#@&($"1
Z$/&"8*V%$$/
LD>=@DG=
^5$$/\&*="%:
.NQON>:'-DS<
9%0/1$&B5%9%0/1$&B5%="%:
Z'8B5%/*="%:
W>99<VD>Q
?"'1"*<"8$
Z'8B5%/
R05()
"GM9>=D<
P9>CTAG9
!'4)*K0"1 !"#@&($"1
;"%J'7:*C3$/5$
."1B%0:$
K0-"8
V%03$
2":
="11'/4(0/PA>K'*D:AG
2A<B9>'&A<9
U14J"%$
K0"1 ?"%-8$B0/$
."('#$% ;$&(B05%/$*
K0"1
="%:
9"-&J"($%
="11'/4(0/
R)$@)$%1\&
95&)
9":$%
U14J"%$
*R(%$$(
."/7"&($%
LD>=@' K0"1
^5$$/&J"%AGC<B
V"($
!088"/1
>0(('/4
90/1*R(%$$(
="%:
!'88*V"($
LD>=@'
$9E='
"?=D<
?"%B8$*C%7) "?=D<
%AGC<B'
Z$/&'/4(0/ ,>DAQSA:
!'4)*R(%$$(
28-#@'"
Z$/&'/4(0/
!-1$*="%:
,0%/$%

Future Directions for the Washington, DC Metro Map

="%:*K0-"8
U"&(*
C7(0/

>0%()*U"8'/4
;$&(*C7(0/
U"8'/4
9%0"1J"-

instead the shape of the complexity has merely been changed, and
the underlying structure of the network is buried just as much as
U"8'/4*,0##0/
before. In usability studies, maps of the same network with more
R05()*U"8'/4
C7(0/
corners tend to be harder to use than maps with fewer corners
[2,3]. *T0J/
>0%()D'$81&

;)'($
,'(-

C7(0/
,$/(%"8

;001*."/$
R05()
C7(0/

R)$@)$%1\&*
95&)
?"%:$(

,)'&J'7:
="%:

90&(0/*?"/0%

9$8&'_$*="%:

.DN=@'2A>>DS
;'88$&1$/*V%$$/

.DN=@'*NCEGCM
;'88$&1$/*A5/7('0/

C8@$%(0/
>0%()*C7(0/

!"/4$%*."/$

$9IOG9:'F9<=>AG
!$"()
V0&@$8
2":
.=D<9O>CQB9'PA>K

6088'&*!'88

*NCEGCM'WA>Q9<E
!"%8$&1$/

R51B5%-*T0J/

=$%'3"8$

>$"&1$/

;?K9<@AI
R(0/$B%'14$*="%:

V081)"J:
K0"1

RJ'&&*,0(("4$
R(G*A0)/\&*;001

?0%/'/4(0/
.=)'(D@<ZE'$DDQ
.DN=@' !'4)B5%-*+
8CGON><'PA>K,"8$10/'"/*K0"1
,%$&7$/(
2AIME=9AQ S&8'/4(0/ 6"8&(0/
A5/7('0/
!ACQA'RAG9
Z'/4\&*,%0&&
W>9A='
$A>SC?K'"T9<N9
R(G*="/7%"&
$9E=ODN><9' *D:AG'
PD>=GA<Q' %NE=D<'
!"44$%&(0/
PA>K

K$4$/(\&
="%:

+AK

&AQO>DK9'
W>DT9

PAQQC<B=D<

U5&(0/*R[5"%$

V%$"(
=0%(8"/1
R(%$$(
%AE='
"?=D<

PAQQC<B=D<

&A=CI9>'
*DAQ;"%%$/*R(%$$(

C/4$8

%QBSA>9' !A>:G9OD<9
*DAQ

,A:ESA=9>
K5&&$88
R[5"%$

V0014$*R(%$$( LD==C<B'
$@C=9'
$DDQ'&A<9
FC=:
2CGG'WA=9

.=>99=

%QBSA>9'*DAQ

U5&(0/

\N99<ESA:

F>9E?9<=
%NE=D<

!0#$%(0/

.[NA>9

8C<B

$A>>9<'
.=>99=

*9B9<=ZE'
,AK9>'!0I(0/
.=>99=
PA>K

281
R(%$$(

!"7:/$
F>D
;'7:
.=)'P

WDDQB9'
.=>99=

]"%%'/410/
!A>OG9'
">?@

.$
?'

!D><C<B=D<'
!"7:/$-*,$/(%"8

*NE
.[N

R)0%$1'(7)

9$()/"8
-D==9<@AI'
FDN>='*DAQ
V%$$/

,D<Q'
+]VD>Q'
.=>99=
?00%4"($FC>?NE

R($@
V%$

,)"/7$%T0(($/)"#*
9"%B'7"/
."/$
,05%(*K0"1 2DGGA<Q'
2DGOD>
&A<?AE=9>'
.'3$%@008*R(%$$(
PA>K
WA=9
FDT9<='WA
;)'($7
!08B0%/ R(G*="58\&
&9C?9E=9>'
PC??AQCGG:'
V"%1$/
FC>?NEC814"($ .[NA>9
9"/:
2CB@'.=>99='
89<EC<B=D<'
WDGQ@ASK'
T0J$% C814"($
2:Q9'PA>K'
*DAQ
98"7:D%'"%&
89<EC<B=D<
^+G:IMCA_
U"&(
!'88
.$'7$&($%
='77"1'88W>99<'
FD><9>
R[5"%$
,'%75&
F@A>C<B'F>DEE
PA>K
?"/&'0/*!05&$
8<CB@=EO>CQB9
T$#@8$
,A>D<E'
?0/5#$/(
T0J$%
%A>GZE'
,"//0/
%IOA<KI9<=
,)"%'/4 FDN>=
2AII9>EIC=@
FDN>=
V"($J"R(%$$(
,%0&&
$9E=IC<E=9>
RC?=D>CA
R)"1J$88
;$&(D$%%
;$&(#'/&($%

2ID0%1
*,'%75&
.@9M@9>QZE'

.@9M@9>QZE'
,NE@
,03$/(

We need to make the distinction between the angle-rule chosen

2&($%8$T5%/)"#
V%$$/*="%:
V8057$&($%
V%$$/
to depict a network, and the quality of implementation using
a given
!"##$%&#'()
K0"1
!05/&80J
R("#D0%1
Z/'4)(&B%'14$
U"&(
9%00:
V5//$%&B5%angle-rule. With skill, a reasonably effective design is likely!05/&80J
R05()
K"3$/&705%(
;$&(
9"%0/&
U"%8\&
<'7(0%'"
,$/(%"8 O%)-,'#"@31*)$,1
Z$/&'/4(0/
="%:
,05%( Z$/&'/4(0/
,05%(
O%)-,'#
irrespective of angle-rule,J./5.%(
but without skill, an unusable
design
is all
.'#$)05&$
;"@@'/4
!05/&80J
U#B"/:#$/(
R80"/$
J',?6/7
R(G
J.,'6/74,,#
;$&(
R[5"%$
A"#$&\&
O%)-,'#"H$&."S)'//)
;$&(*9%0#@(0/
="%:
too likely. Fundamentally,
we do not have
rules for arbitrary
reasons,
Z$J*V"%1$/&
A$*N(%154,').
!(/'5.%( J.%6-,1)"
!"((0/*,%0&&
R05()J"%:
9$%#0/1&$- K0()$%)'()$
I"0%)$(/'"
;$&(*S/1'"
A poorly designed map with many ='#8'70
corners disrupting
the lines (left)
will be
rules exist so that objectives may
be attained. The objective
in M35./7
;"($%800
^5".0/10/
."#B$()
!$"()%0J
,"/"1"*;"($%
9%'14$
>0%()
U.$5"5/*)$,1"$5"
T$%#'/"8&*
]58)"#*9%0"1J"K'7)#0/1
,"/"%-*;)"%D
H$&."M%'1/)
harder to use than a map of the same network
with fewer
MF*PF*Lengaging,
J,*N-,5)/'5corners (right).
1,)"5+%)$%667"
producing a transport schematic
is
an
attractive,
simple,
R5%%$-*^5"-&
90%054)
>,,'"Q%'N
*,1-$&3'/#
!$%0/*^5"-&
="%&0/&*V%$$/
A designer must
always try to optimize<"5I)"88
geometry:
are the lines as straight
U,))/'$#&/"I"
usable design that encourages people to travel and enables
to
J%'+/1#/'5"Q%'N
R05()*^5"!$"()%0J them
S#@$%'"8*
Z$//'/4(0/ U8$@)"/(
O./)5),1/
K%N4,,#
23"8
T$%#'/"8*Y !$"()%0J
+*,"&(8$
,%0&&)"%B05%
;)"%D
as
possible,
with
as
few
corners
as
possible?
=5(/$-*9%'14$
T$%#'/"8*X
R(07:J$88
plan quickly without making mistakes. Adopting the traditional
?517)5($
>$J*
,8"@)"#
S)%1(,'/
S&8"/1*V"%1$/&
,%0&&*
A5/7('0/
9%'I(0/
V"($
>$J
octolinear angle-rule provides a tool by which this objective
may be
R,').4,,#
U"&(*=5(/$,8"@)"#*>0%()
,%0&&
W#&4%'/
S,3).&%)/
H%)*."W1#
,8"@)"#*,0##0/
O,,#5$#/"Q%'N
9%07:8$,
reached, but if the tool is used inappropriately then the
designer
will
J%1,15"Q%'N
R,').4,,#"H$665
,8"@)"#*R05()
balancing
act, andR05()D'$81&
it >$66"H$66"W%5)
is worth considering
whether
topographical
J,6$1#%6/
!0/0%*2":
H/%#5),1/"0%1/
V%$$/
O/5)"]$1*.6/7
!'1,5"X',:/
9"8)"#
fail to attain the objective. Conversely, given the objective of an
]0%$&(*!'88
6$@(D0%
information on
a;'#B8$10/*="%:
journey planner
does serve a usefulR-1$/)"#
purpose.
]$1*.6/7"J/1)'%6
M3'1)"K%N
H%'',4"I"
U83$%&0/*K0
T00('/4*9$7
Q$11/'
M,31#5"X'//1
.$J'&)"#
effective map, there may be other tools available to enable
this to beO/%6#5),1/
^3//1583'7
K'3$%*T)"#$&
T00('/4*9%0"1J",%-&("8*="8"7$
;'#B8$10/
,088'$%&*;001
89:
met. As shown in Appendix B, octolinear is just one of manyR,')."H%'',4
possible
!"#$%&'(&'$)
R05()*;'#B8$10/
H/1#,1"J/1)'%6
Identification
or destination
stations. Technically,
the
H%'',4Y,1Y
[$1&583'7
!9=>DMDGC=A<'&C<9
,AK9>GDD'&C<9
HD?KGA<QE'&CB@='*ACGSA: of start
O,,#"X'//1
W%5)"]$1*.6/7
)./YH$66
=$/4$*;$&(
[/1),1
?0%1$/
F9<=>AG'&C<9
2AII9>EIC=@'J'FC=:'&C<9
LD>=@9><'&C<9
A%71/'5"
A3$56$+"
angle-rules, and different angle-rules are suited
to
different
9":$%800*.'/$
!"##$%&#'()*+*,'(-*.'/
C/$%8$0%1/
A3$56$+ >%1,' W%5)*,)/
PC??AQCGG:'&C<9
FC>?G9'&C<9
(NOCG99'&C<9
M'/1)"J',55
purpose
of
a
journey-planner
is
to
assist
the
user
who
already
O/5)"
R,').4$*N"
U3'1+$N/"0%1/
Q'/5),1"
A5B'8$$*.'/$
,$/(%"8*.'/$
RC?=D>CA'&C<9
HCE=>C?='&C<9
&D<QD<'+T9>B>DN<Q
H%'',4
Q%'N
O/5)"A3$56$+
A,%#
networks.
>0%J001*A5/('0/
$A=9>GDD'J'FC=:'&C<9
?$(%0@08'("/*.'/$
,'%78$*.'/$
knows the start and
destination stations. Typically, tourist
29A=@>DS'
-9>IC<AGE' 2A==D<'
'0/'5/'X
F>DEE

29A=@>DS'
-9>IC<AG'Y

"?=D<'
F9<=>AG

%AGC<B'FDIID<

.DN=@'
"?=D<

2DN<EGDS'
F9<=>AG
+E=9>G9: LD>=@VC9GQE "?=D<'
-DS<

2DN<EGDS' 2DN<EGDS'
%AE=
$9E=

,DE=D<' .DN=@'
!A<D> %AGC<B

F@CESC?K'
PA>K

29A=@>DS'
-9>IC<AG'1

-N><@AI'
W>99<

,NE@'!A>K9=

.=AIVD>Q' *AT9<E?DN>='
,>DDK
PA>K

WGDN?9E=9>'
.DN=@'
*DAQ'
89<EC<B=D<

$9E='
89<EC<B=D<

.GDA<9'
.[NA>9

.=)'(AI9EZE'
PA>K

$9E=',>DIM=D<

WN<<9>EON>:

PCIGC?D

89S'WA>Q9<E

UNG@AI',>DAQSA:

*C?@ID<Q

PA>ED<E'W>99<

;IM9>CAG'
$@A>V

RAN]@AGG

PN=<9:',>CQB9

.=D?KS9GG

FGAM@AI'LD>=@

%AE='PN=<9:

FGAM@AI'FDIID<

FGAM@AI'(N<?=CD<

D0

FGAM@AI'.DN=@

.DN=@VC9GQE

,AG@AI

$CIOG9QD<'PA>K

$CIOG9QD<

<

2CB@',A><9=

T*N/1.%(

FA>M9<Q9>E'PA>K

'2A>>DS

A>>DS'
=@9a2CGG

A>>DS

%QBSA>9

.=A<ID>9

FA<D<E'PA>K

29AQE=D<9'&A<9

LD>=@SC?K'
PA>K

LD>SDDQ'(N<?=CD<

.DN=@'$CIOG9QD<

-@9:QD<',DCE
X,6#/'5"X'//1

.DN=@'89<=D<

H$&.&%)/
;$&(*,%0-10/
J',3*."
H$66

!D>Q9<

$9E='F>D:QD<
>0%()$%/*.'/$
U,))/1.%("
H%6/

.=D<9O>CQB9'PA>K
2A>G9EQ9<

,>D<Q9EON>:'
PA>K

8CGON><

UC<?@G9:'*DAQ'
J'U>DB<AG

2AIME=9AQ
2AIME=9AQ'
29A=@

-N><MCK9'&A<9

F>DN?@'
2CGG

">?@SA:

WDEM9G'
+AK

3MM9>'2DGGDSA:

-NV<9GG'PA>K

6'&(%'7(*.'/$

&DNB@=D<

$DDQVD>Q

$DDQ'W>99<

2CB@BA=9

WDGQ9>E'W>99<

$CGG9EQ9<'W>99<

H9OQ9<

F@CBS9GG

,DN<QE'W>99<

%AE='UC<?@G9:

,>9<='F>DEE

HDGGCE'2CGG

$9IOG9:'F9<=>AG

;<=9>?@A<B9'.=A=CD<E

"><DE'W>DT9

UC<?@G9:'F9<=>AG

L9AEQ9<

LD>=@'$9IOG9:

O/(86/7"Q%'N

R,')."
O/(86/7

.DN=@BA=9

,N?K@N>E='2CGG

29<QD<'F9<=>AG

$9IOG9:'PA>K

$DDQECQ9'PA>K
$9E='
UC<?@G9:

FDGC<QAG9

8C<BEON>:

P>9E=D<'
*DAQ

-D==9>CQB9'J'$@9=E=D<9
S,3)."H%'',4

A3$56$+"X%'#/15
!CGG'2CGG'%AE=

H$66$1&#,1

,N><='+AK

\N99<EON>:

2A>>DS'J'$9AGQE=D<9
89<=D<

"<9>

S/:/1"
attractions and other important destinations in X'//1"0%1/5
most
will
H%''$1&%7" cities S$5)/'5
M6%*N.,'5/"A
advertise their most convenient stations in their publicity. S,3)." R("('0/
S,3)."A3$56$+
2?8'$#&/
U,))/1.%(
!'*.4%7
S3#83'7"H$66
R/%5#/1
X,5+/6"
!"#$%&&'()'*+,%*-./'01-2'(3&4/'5667
C*#"@*0D*()$*.0/10/*E/1$%4%05/1*()"(*0B$-&*()$*1$&'4/*%58
>%1,'"H,35/
Users
familiar with their
their
choice
of
O%6
H%(+5)/%#locale
]$1583'7"
K%N will know
L,66$5"H$66
2++/'"
]$1*.6/7"
Q%'N
J/1)'%6
H,66,4%7
The goal of a designer is to optimize the geometry
ofS3#83'7"U,41
a map by O/(86/7"
O/5)" A,%#"I"
X'//1-,'#
O%6).%(5),4"
!'5/1%6
starting
For
people
either
of
these,
H%(+5)/%#" who do not know
O$66/5#/1" stations.
H%(+5)/%# ]',&1%6
^3//15"A,%#
H/%).
X'//1 [$683'1
U3-1/66"Q%'N
S),1/8'$#&/"Q%'N
converting the complexities of reality into simple straight lines.
]$1*.6/7"
and who do not possess
a scale
streetmap,
it is unlikely that a
A,%# M/65$_/"
H,66,4%7"
R,').,6)
[/1)$5."
[/1)$5."U,41
A,%#
Q%'N
M',1#/583'7
However, in opposition to this is the perceived, or real need to!6+/'),1
limit
U,41"
J%6/#,1$%1"
O/5)
schematic S,3)."H%(+5)/%#
map can S4$55"
assist.
Identifying
key
stations by their
M',1#/583'7"Q%'N
J,))%&/
H%'6/5#/1
J.%6N"
A,%#
"0/7),1">$#
Q/'$:%6/
]%'(
[/15%6"
topographical distortion. Optimizing the two can be a very
difficult
O$66/5#/1"
J%(#/1"A,%#
A$5/
position
on
a
relatively
small-scale
schematic
that isJ%1,183'7
bereft of
@31*)$,1
^3//1V5"
J%(#/1"
[$683'1"

Preservation of topography

2A=?@'%<Q

FD?KVDE=9>E
S,3)."
[/1),1
+AKSDDQ

-DD=C<B',>DAQSA:

FDGGC9>E'$DDQ

%MMC<B

CB@'.=>99=

-DD=C<B',9?

.9T9<'
.CE=9>E

2A>>C<BA:'
W>99<'&A<9E

-D==9<@AI'
2AG9

,GA?K@D>E9' $AG=@AIE=DS'
*DAQ
F9<=>AG

.DN=@'$DDQVD>Q

.DN=@'
-D==9<@AI

$A<E=9AQ

$AG=@AIE=DS'
\N99<E'*DAQ

2AC<ANG=

UAC>GDM

.<A>9EO>DDK

!A<D>'2DNE9

W>A<B9'2CGG

*DQC<B'
RAGG9:

,A>KC<BECQ9
WA<=E'2CGG

&9:=D<E=D<9
Q%'N
L9SON>:'
*9QO>CQB9
L%65),1"[$1&56%1#
U,41
H$&."A,%#
PA>K
J%6/#,1$%1"A,%#"
H$&.83'7"I"
S)<"@,.1V5"O,,#
>,'1$1&),1"
I"M%'1583'7
T56$1&),1
L%65),1"
&9:=D<E=D<9'
&9:=D<
[$683'1"Q%'N
[/15%6"X'//1
J'/5*/1)
.=>A=VD>Q'
89<EAG'W>99<
@31*)$,1
2CB@'*DAQ
H,(/'),1
;<=9><A=CD<AG
8CGON><'
>%$#%"P%6/
\N99<ZE'PA>K 2CB@'*DAQ
.SCEE'FD==AB9
.=>A=VD>Q
DS<
$A<E=9AQ'
H%*N1/7"
[$1&V5"J',55"
R,')."
PA>K
W35),1
R,')."
J/1)'%6
S)<"Q%1*'%5
.=)'(D@<ZE'$DDQ
.DN=@'
2DI9>=D< 2A?K<9:'
FAG9QD<CA<' 2CB@ON>:'J'
!D><C<B=D<'
!1&/6
H%&&/'5),1
8CGON><'PA>K
W%6$1& 2A?K<9:' !*),1"
X'/%)"
M%N/'"
2AIME=9AQ
*DAQ'J'
HAGE=D<'
;EGC<B=D< W%6$1&"
F>9E?9<=
$C?K
F9<=>AG
$DDQB>A<B9'
O/5)8,3'1/" O%'4$*N"
O/5)"
,A><EON>:
(N<?=CD<
>%'76/8,1/
S)'//) Q,')6%1#"
!ACQA'RAG9
PA>K
.=>A=VD>Q'
M',%#4%7
Q%'N"
!:/13/
S)'//)
!*),1
2ABB9>E=D<
W%5)"
2CB@'.=>99=
H%*N1/7"O$*N
%NE=D<
W>9A='
+GQ'
W#&4%'/"
!*),1
PNQQC<B'
$9E=ODN><9' *D:AG' $A>SC?K'"T9<N9
PD>=GA<Q' %NE=D<'
.=>99=
A,%#
2D]=D<
"OO9:'*DAQ
W35),1"S\3%'/
.=>99= .[NA>9
PA>K
+AK
PAQQC<B=D<
G
!CGG'&A<9
%QBSA>9'*DAQ
!CG9'
W%6$1&"J,((,1
A,7%6"
Q%##$1&),1
A/&/1)V5"
H,?),1
,9=@<AG' %<Q
%AE='
"<B9G
,DS'
O%''/1"S)'//)
0%#8',N/"
%GI'
3MIC<E=9>'
HAB9<@AI'
K%N
.@D>9QC=?@
,DS'
A355/66"
W>99<
Q%'N
W#&4%'/"
*DAQ F@N>?@
PGACE=DS
2AI
X',:/
PA>K
,>CQB9
29A=@SA:
,A>OC?A<
3M<9:
PAQQC<B=D<
X,,#&/"S)'//) S\3%'/
8C<BZE'
!*),1"
$A>>9<'
!DD>BA=9
S,3)."
A,%#
*9B9<=ZE'
&AQO>DK9'
%QBSA>9' !A>:G9OD<9 ,AK9>'
K6#"S)'//)
.=>99=
F>DEE'
J/1)'%6
O.$)/"J$)7
R,').-$/6#5 W%6$1&
0%)$(/'"A,%#
W>DT9
.=>99=
PA>K
*DAQ
M%754%)/',9?D<=>99 HAB9<@AI' 2D><?@N>?@ 3MIC<E=9>
.=)'PA<?>AE
]%''$1&#,1
&CT9>MDDG'
3M=D<' ,A>KC<B
O,,#"0%1/ ,>DIG9:a'
U,))/1.%("
$9E='
!*),1"
UA>>C<BQD<
.=>99=
Q%##$1&),1
O:a,DS
%AE=
PA>K
M/).1%6"
J.%1*/'7"
M,1#" K?-,'#"
J,3')"
R,))$1&"
.=9M<9:'
2AI
M,5),1">%1,'
U,41
,A:ESA=9>
WDDQB9'
*NEE9GG'
&A=CI9>'
M%'8$*%1
H9TD<E'*DAQ
"GQBA=9'
>$6/"W1#"
X'//1
0%1/
W>99<
S)'//) J$'*35
A,%#"
H$66"X%)/
S./+./'#V5"M35."
S.,'/#$)*.
.[NA>9
*DAQ
.=>99=
H/%).',4"
%AE=
.=A>'&A<9
$@C=9?@AM9G >%'N/)&A<BQD<'PA>K
>,,'&%)/
U/'($1%6"Z
"GQBA=9
H/%).',4"
H,68,'1
S,3)."!*),1
>%'86/"!'*.
^3//154%7
0%1*%5)/'"
H,66%1#"
S./+./'#V5"
%AE='
LD>=@'
F@A<?9>:'
$@C=9'
.=)'
!A>OG9'
='
U/'($1%65"
,A<K
LD==C<B'
-D==9<@AI'
,D<Q'
+]VD>Q'
0/$*/5)/'"
X%)/
Q%'N
M35.
$DDQ'&A<9
"GG'.AC<=E
-DS9>'
"?=D<
S)<"Q%36V5
"?=D<
\N99<ESA:
&A<9
FC=:
PANGZE
">?@
<
X,6#.%4N"
S)/+1/7"X'//1
M,
FA<<C<B'
FDN>='*DAQ
.=>99=
FC>?NE
D9"E9"`
2CGG'WA=9
-DS9>'2CGG
0$:/'+,,6"
S\3%'/
WA=9SA:
-DS<
A,%#,GA?KSAGG
A,
&CI9@DNE9
J.$54$*N"Q%'N
S)'//)
,A<K
M%1N
O.$)/*.%+/6
M6%*N-'$%'5
J,:/1)"X%'#/1
2DGOD>< K5)/'6/7 !A<ECD<'2DNE9
P>C<?9' ,9?K=D<' WAGGCD<E'
*D:AG'
2DGGA<Q'
.@9M@9>QZE'
&A<?AE=9>'
.@9M@9>QZE'
[/15$1&),1"
U3'1.%("
PA>K
WA=9
PA>K
*9B9<=
RC?=D>CA
*9A?@
Q$**%#$667"
,NE@
FDT9<='WA>Q9<'
,NE@'!A>K9=
PDMGA> %AE=' K67(+$%
H$&."S)'//)"
$9E=V9>>:
X311/'583'7 .@AQS9GG X'//1 S)%(-,'#"M',,N
!6#&%)/
H,3156,4"W%5)
J$'*35
&9C?9E=9>'
[/15$1&),1
X'//1"Q%'N
PC??AQCGG:'
;<QCA
!D<NI9<=
!6#&%)/"W%5)
FA<<D<'
U/(+6/
.[NA>9
FC>?NE
>%15$,1"H,35/
A%:/15*,3')"
F:M>NE[1$&.)58'$#&/
,9?K=D<
FNE=DI' *D:AG'
.=>99=
2CB@'.=>99='
89<EC<B=D<'
WDGQ@ASK'
"?=D<'
U,4/'"X%)/4%7
,GA?KV>CA>E
Q%'N
"GO9>=
2DNE9
2:Q9'PA>K'
*DAQ
F9<=>AG
J%11,1"S)'//)
89<EC<B=D<
^+G:IMCA_
0$(/.,35/
J.%'$1&"J',55
$AMMC<B
H/%).',4"
H,3156,4"
H,3156,4"
H%)),1"
W(8%1N(/1)
S,3)."
W>99<'
FD><9>
$9E=';<QCA'
M%',15"
W%'6V5" $9E='.CGT9>=DS<
U,4/'"
>,13(/1)
U/'($1%6"B
J/1)'%6
J',55F@A>C<B'F>DEEO/5)
H7#/"Q%'N"
PA>K
S.%#4/66
[/15$1&),1
-9IMG9
\NA:
J,3')
J,3')
0,1#,1"
H%((/'5($).
H$66
.DN=@'
J,'1/'
8<CB@=EO>CQB9
PD<=DD<'HD?K
O%)/'6,,
M'$#&/
,A>D<E'
%A>GZE'
LD>=@'
%IOA<KI9<=
[/4"X%'#/15
"?=D<
Q,+6%'
2AII9>EIC=@ FDN>=
FDN>=
W>99<SC?@
*D=@9>@C=@9
&D<QD<'FC=:'"C>MD>=
$9E=IC<E=9>
O/5)-/''7
$A=9>GDD
RC?=D>CA
FA<A>:'
S,3).4%'N
O/5)($15)/'
&D<QD<'
X6,3*/5)/'"
O/5)"
$@A>V
FA<AQA'$A=9>
,>CQB9
O%++$1&
8C<B'W9D>B9'R
O/5)"T1#$%"^3%7
A,%#
[/15$1&),1
WGDN?9E=9>'
.DN=@'
FA<A>:'
$9E='
.GDA<9'
-N><@AI' .=AIVD>Q' *AT9<E?DN>='
.=)'(AI9EZE'
S6,%1/" P$*),'$%
S)<"@%(/5V5"
J%1%'7"O.%'0%(8/)."
*DAQ'
89<EC<B=D<
29>D<'\NA:E
,9>ID<QE9:
89<EC<B=D<
.DN=@SA>K A$*.(,1#
$@A>V
.[NA>9
W>99<
,>DDK
PA>K
PA>K
M,',3&.
S\3%'/
Q%'N
O/5)"M',(+),1
R,').
H/',1"^3%75
A,)./'.$)./
.DN=@'\NA:
&AIO9=@'
J%
$9E=',>DIM=D<
LD>=@
]36.%("M',%#4%7
$DDGSC?@'">E9<AG
S,3)."^3%7
<9>EON>:
.N>>9:'\NA:E
F>DEE@A>ODN>
,D>DNB@
PCIGC?D
M/'(,1#5/7
Q$(6$*,
!NQ?@N=9
W6/+.%1)"I"J%5)6/
J%1%#%"O%)/'
Q%'5,15"X'//1
J',55.%'8,3'
%G9M@A<='J'FAE=G9
T(+/'$%6"
;EGA<Q'WA>Q9<E
E
UNG@AI',>DAQSA:
89<<C<B=D<
O.%'[/11$1&),1
S3''/7"^3%75

>DS

CGG

89<EAG'*CE9 ,>D<Q9EON>:
$CGG9EQ9<'(N<?=CD<

$9E='2AIME=9AQ

UC<?@G9:'*DAQ

UC<EON>:'PA>K

89<=CE@'
">E9<AG
89<=CE@'-DS<
-DS<'
$9E=
FAIQ9<' 2DGGDSA:'*DAQ
*DAQ
F@AGK'UA>I
FAIQ9<'
FAG9QD<CA<'*DAQ
-DS<

&9:=D<'!CQGA<Q'*DAQ

,9GEC`9'PA>K

H%1&/'"0%1/
Q%'N"A,7%6
HAGE=D<'
FA<D<ON>: 8C<BEGA<Q

Below: London Underground map, optimized for geometrical simplicity.

Right: London Underground map optimized for topographical accuracy.

Page 16

PA>ED<E'W>99<

;IM9>CAG'
$@A>V

L9S'F>DEE'WA=9

RAN]@AGG

,>D?KG9:
PN=<9:',>CQB9

.=D?KS9GG

Q3)1/7"M'$#&/

L9S'F>DEE

2D<D>'+AK'PA>K

W>99<SC?@
H9M=VD>Q',>CQB9
%GT9>ED<'*DAQ

>3#*.3)/

P%3?.%66

FN==:'.A>K

+TAG

K:%6

T56%1#"X%'#/15

W%5)"Q3)1/7
J6%+.%("

S),*N4/66

J3))7"S

Future Directions for the Washington, DC Metro Map

street and other details is a skill that could only be displayed

by a user with considerable city expertise.
Identification of alternative stations. At times of service
disruption, topographical information on a schematic might be
useful for identifying a nearby alternative station that could be
used instead, although announcements, signs and
neighborhood maps may fulfill this function more effectively.
Disambiguation of similarly-named stations. Where networks
name stations after intersecting streets, topographical
information on a schematic provides a way of disambiguating
them. For example, with five subway stations named 23rd
Street in Manhattan, an accurate topographical map might
help the user to identify the target station. The Chicago El is
another example, with four stations named Pulaski.
Identifying and locating between-station transfers. For older
networks, originally constructed or operated by competing
companies, potential transfers between rivals were given little
consideration. In many cities, London included, certain
journeys may be completed faster by leaving a train at one
station, and walking a few hundred yards to another station for
a new train. Some of these transfers now have official status,
and are recommended on the London Underground map. The
introduction of electronic ticketing has led to others that are
not officially recommended. Even so, novice users may be
reluctant to venture into unknown neighborhoods, and without
clear signposting, they may have difficulty locating even
recommended between-station transfers.
Minimizing conflict with users mental models of the city. Some
users have very low levels of distortion-tolerance: any
topographical disruption will lead to complaints, although it is
not clear that this will damage their journey planning. Indeed,
the people who tend to complain the most are users who are
the most expert, and therefore have least need of topographical
maps. Novice users are unlikely to notice even quite high levels
of distortion, especially in suburbs.

The above points indicate that topographical information, even

on a schematic map, may be of some use, although its importance is
probably overstated. If such information is to guide the configuration
of a map, it is important that it is consistently applied. This is
because, from the point of view of a novice user, in the absence of
further information, it is impossible to know which configurations on
a map constitute travel hints, and which are unintended byproducts
of the schematization process. Even where topography is impeccably
accurate, such information is not always usable. For example,
consider one station that is exactly north of another by a few blocks.
This information can only be used for navigation if the direction that
the station exit is facing is known, otherwise, a compass will be
required.
Overall, although schematic maps can be configured to show
topographical information precisely, this can be at the expense of an
attractive usable diagram and, in any case, it is not clear that the
user will be able to take advantage of such information without a
scale streetmap for additional guidance. In which case the streetmap
should be consulted for full guidance rather than the schematic.
From the point of view of the designer, the best strategy is to
optimize geometry on the schematic as much as possible, taking
advantage of topographical distortion in order to do so, but not
distorting reality to the point at which typical users are taken out of
their comfort zones. Certainly, it is necessary to avoid conveying any
information that could be misleading if taken literally by a user.
Conclusions
Blind adherence to traditional design rules is unlikely to lead
to a usable map. The designer must take care to optimize its
geometry, ensuring that reality has genuinely been simplified. The
aim is to take the complex line trajectories, and convert them, where
possible, into memorable, easy to follow straight lines. Changes of
direction corners should be minimized: nothing will be gained if
complex twisting curves are converted into complex twisting zigzags.
In the process of optimizing geometry, some topographical distortion
is inevitable. Although complaints about topographical distortion are
probably overstated, the designer must have some sensitivity of the
issues so as not to go beyond people's comfort zones.
Page 17

Future Directions for the Washington, DC Metro Map

Appendix B: Taxonomy of schematic angle-rules

There are many ways in which to design a schematic map. The
traditional way, an octolinear map, with four possible angles
(horizontal, vertical, 45 left to right, 45 right to left) is just one of a
family of angle-rules that could be adopted, and have been adopted
round the world by various transport undertakings. What is notable
is that different networks have different structures, and that these
are suited to different angle-rules [2,3].

Examples
The current Madrid Metro map. Many designers seek to use as few
diagonals as possible. For example, in the 1970s the Vignelli New
York Subway map was effectively tetralinear in Manhattan.

The basic categories are outlined in this section, along with

brief notes on their advantages and disadvantages. It is difficult to
know in advance which angle-rules will suit which networks because
this research is still in its infancy. Only the simplest cases are
named, but this list provides a useful starting point for any designer,
irrespective of network.
This analysis has recent origins and, unfortunately, looks set to
inherit mathematical terminology. The angle combinations available
for a particular map are categorized by their linearity. Imagine a
design that will have just vertical and horizontal lines (two angles).
With a blank sheet, the designer has a choice of four directions in
which to draw a line (up/down/left/right). Hence, a two-angle map is
tetralinear. The traditional schematic angle-rule (horizontal/vertical
lines and 45 diagonals) gives the designer a choice of eight different
directions, hence the map is four-angle or octolinear, and so on.
B.1 Tetralinear/Two angles

Version a:

Version b:

Above: Current Madrid

Above right: New York, 1972

Comment
Too severe for most
cities, both of the above
maps are controversial,
and the Madrid map
demonstrates the
difficulty in configuring
transfers with this type
of design. However, a
rotated tetralinear map
can be very striking, and
is essentially the design
used by Montreal (right).

Advantages
Relatively compact
Disadvantages
Can lead to excessive topographical distortion, or many corners,
sometimes both.
Page 18

Future Directions for the Washington, DC Metro Map

B.2 Hexalinear/Three angles

B.3 Octolinear/Four angles

X00(),

K1/J216.$*6
L(,#"I&*)$.

Version a:

V#$7+1)"I1(6

P&JB11+

Version b:

V1..$*(+,$"M"O#$.6.1)$
N$E+$)

R15.#,&.$

O11+6(+$"T&*J
Advantages
O$6."W()/#'$7
Fits many networks surprisingly well. Version (a) is inherently !*)16"Y*1Q$
R.&)-1*$
compact horizontally.
60 angles can give a dynamic
appearanceI15)+6"Y*$$)
to a
W()/#'$7"K$).*&'
X+,B&*$
map.
K&)1)6"T&*J
O11+"Y*$$)
X&6."W()/#'$7
I5*)."P&J
A(''"L(''"X&6.

K1'()+&'$
a5$$)6E5*7
Disadvantages
L(,#,&.$
V5*)0(J$"3&)$
L$)+1)"K$).*&'
More
restrictive
than
octolinear,
the
designer
must
abandon
either
K*15/#"
L&**(),&7"
^(),6E5*7
!*/#B&7
L(''
Y*$$)"3&)$6
I*$)."K*166 lines.
horizontal or vertical
400$*"L1''1B&7

1)$

T*$6.1)"
D1&+

*J

O$-E'$7"T&*J

A&)1*"L156$

Y1'+$*6"Y*$$)

Examples
L&-06.$&+
N1''(6"L(''
None currently
in use.
U$&6+$)

L&-06.$&+"
L$&.#

V52)$''"T&*J

D1+(),"=&''$7

R$Q$)"
R(6.$*6
R15.#"
V1..$)#&-

Y*&),$"L(''

Disadvantages O11+21*+
L&()&5'.
V1..$)#&-" I'&/J#1*6$" O&'.#&-6.1B"
R15.#"
D1&+
L&'$ Frequently
K$).*&'
abused
by careless designers. 45 diagonals are
O11+21*+
W&(*'10
inherently wasteful of space.
I&*J(),6(+$

R)&*$6E*11J

O&'.#&-6.1B"
a5$$)6"D1&+

Examples
Y&).6"L(''
O&)6.$&+
Approximately
1/3 to
1/2 or
world urban rail maps are octolinear [1].
3$7.1)"A(+'&)+"D1&+

W()6E5*7"T&*J

Y160$'"
P&J

O(''$6+$)"Y*$$)

Advantages
315,#.1)
Ostensibly a design gold standard. Well-known and well-understood
K#(,B$'' and flexible. Computer software
I5/J#5*6."L(''
by designers worldwide.
Versatile
presets make this easy to implement.

D$+E*(+,$

U$BE5*7"T&*J

3$7.1)6.1)$

!*6$)&'

^$).(6#"V1B)
W()/#'$7" I$'6(8$"
3$7.1)6.1)$"L(,#"D1&+
Comment^('E5*)
Comment
^$).(6#"
D1&+"M"
T&*J
V1B)"
W*1,)&'
L1''1B&7"D1&+
A very
subtle
style, unlikely
conflict with users
expectations
The first choice
for
many designers, but often misunderstood. The
I*1)+$6E5*7
O$6."L&-06.$&+
K#&'J"W&*-toO$6.
3$7.1)
R.*&.21*+"
^$)6&'"D(6$
K&-+$)"
S).$*)&.(1)&'
D1&+
regarding map
design.
For
some
networks,
such
as
the
London
Paris
Metro
(below)
is a good example40-()6.$*
where the complexity of the
W()/#'$7"D1&+
K&'$+1)(&)"D1&+
L(,#E5*7"
L&/J)$7"
L&/J)$7"
K&-+$)"
O&)6.$&+"T&*J
^$)6&'"Y*$$)
V1B)
M"S6'(),.1)
K&)1)E5*7
K$).*&'
O(/J
Underground,
it
can
result
in
pleasing
geometric
effects
such
as
network
means
map
is hard-pressed to offer any
R.*&.21*+that an octolinear
40-()6.$*"I*(+,$
R15.#"
a5$$)_6" ^('E5*)"
RB(66"K1..&,$
L&-06.$&+
A1*)(),.1)"
K&'$+1)(&)"
N&'6.1)"
L1-$*.1)
T&*J L(,#"D1&+
L1*)/#5*/#
R.*&.21*+"
equilateral
triangles (below).
benefits over a
topographical
map,
Tokyo
is another.
N&'6.1)" ^(),6'&)+
D1&+"M"
K*$6/$).
L(,#"R.*$$.
I*1)+$6E5*7"
T&*J

^('E5*)"T&*J

R.<"C1#)_6"O11+

A&(+&"=&'$
X+,B&*$"
D1&+

O&*B(/J"!Q$)5$

Y*$&."
T1*.'&)+"
R.*$$.

I&J$*"
R.*$$.

X56.1)

X+,B&*$"
D1&+

O$6.E15*)$"T&*J

D1&+

(.7

3&+E*1J$"Y*1Q$
L1''&)+"
T&*J"

Y1'+#&BJ"D1&+

D$,$)._6"
T&*J

3&)/&6.$*"
Y&.$

^$)6(),.1)"
bP'7-0(&c

I&*1)6"
--$*6-(.# K15*.

L(,#"R.*$$."
^$)6(),.1)

X&*'_6"
K15*.

L7+$"
T&*J"K1*)$*

I&*E(/&)

A11*,&.$

L1'E1*)
K1Q$)."Y&*+$)
3$(/$6.$*"Rd5&*$

V1..$)#&-"
K15*."
D1&+

Y*$$)"T&*J
T(//&+(''7"
K(*/56
R.<"
C&-$6_6"
T&*J

R'1&)$"
Rd5&*$

R15.#"
^$)6(),.1)

V$-0'$
K#&*(),"
K*166

O$6.-()6.$*

!'+,&.$"
X&6.

I1B" I*1-'$7"
K#5*/# `E7`I1B

R.$0)$7"Y*$$)

A1)5-$).

R#&+B$''

O&00(),

*J

T1).11)"N1/J

U1*.#"
Y*$$)B(/#

O$6."S)+(&"a5&7

K&)&*7"
O#&*2

D1.#$*#(.#$
R15.#B&*J

I$*-1)+6$7
31)+1)"
I*(+,$

31)+1)"K(.7"!(*01*.

3&-E$.#"
U1*.#

R15.#"a5&7
A5+/#5.$

I1*15,#

^$))(),.1)
I*1/J'$7

L1)1*"P&J"T&*J

K'&0#&-"U1*.#
K'&0#&-"K1--1)
K'&0#&-"R15.#

U$B"K*166

K5..7"R&*J
Y*$$)B(/#
N$0.21*+"I*(+,$

PQ&'
I*(%.1)

S6'&)+"Y&*+$)6

R5**$7"a5&76

U$B"K*166"Y&.$

R.1/JB$''

O11'B(/#"!*6$)&'

K*166#&*E15*

X'$0#&)."M"K&6.'$
=&5%#&''

^(),"Y$1*,$"=

K&)&*7"O#&*2
L$*1)"a5&76

K&)&+&"O&.$*

X&6."T5.)$7

#2($'+6

K56.1-" D17&'" K70*56 I$/J.1)

L156$ !'E$*.
O$6."R('Q$*.1B)

I'&/JB&'' K&))(),"
V1B)

X-E&)J-$).

T(-'(/1

K'&0#&-"
C5)/.(1)

D17&'" T*()/$" I$/J.1)"Y&''(1)6"

=(/.1*(& D$,$). T&*J
D$&/#

X&6."
S)+(&

T10'&*

O$6.2$**7

I&*J(),

X&6."
L&R.&*"3&)$

3&),+1)"T&*J

3(-$#156$

I$/1).*$$
40)$7

T'&(6.1B

!''"R&().6

V1B$*"
Y&.$B&7

N&,$)#&-"L$&.#B&7
40.1)"
T&*J

O$6."
L&-

I1B"D1&+
N$Q1)6"D1&+

O#(.$/#&0$'

A&)6(1)"
L156$

T&*61)6"Y*$$)

T5.)$7"I*(+,$

K&))1)"
R.*$$.

O&.$*'11
S-0$*(&'"
O#&*2

!'+,&.$
V1B$*"
L(''

I&)J

I'&/J2*(&*6

=(/.1*(&

O$6."
^$)6(),.1)
O$6."I*1-0.1)
W5'#&-"I*1&+B&7

D566$''"
Y11+,$" Rd5&*$
R.*$$.

K#&)/$*7"
R."T&5'_6
3&)$

P%21*+"
K(*/56

A&*E'$"
!*/#

^)(,#.6E*(+,$
Y'15/$6.$*"
D1&+"

O&**$)"
R.*$$.

A('$"
X)+

I$.#)&'"
Y*$$)

R#1*$+(./#

O11+,*&),$"
X'-"T&*J
T&*J
N&,$)#&-"X&6.

!EE$7"
D1&+

T5++(),"
A(''"3&)$

P'+"R.*$$.
3(Q$*011'"
R.*$$.

W&**(),+1)

I1)+"
R.*$$.

a5$$)6B&7

R#$0#$*+_6"
I56#"

R#$0#$*+_6"
I56#"A&*J$.

X56.1)"
Rd5&*$

A&*7'$E1)$

I&76B&.$*
U1..(),"
L(''"Y&.$

L1%.1)
!),$'

T&++(),.1)
D17&'"P&J

C5)/.(1)
L&,,$*6.1)

I&*)6E5*7

^(),_6"
K*166"
R.<"T&)/*&6

W1*$6."L(''

R7+$)#&-

X'Q$*61)"D1&+
3$B(6#&-

Page 19

^,-%8E&#R

Y$8/"D&,8',0
D&,8',0

T,'',-%)*4BE#,)%$

D&,8',0"
D(R-$#8" Y$8/"
A(-*# _(8/+*/$ 3(-$
T(##*5

T(##*5"
*-b/2$bT,''

Z*#/25,+Q"
W(#Q

S#$-/"V#*88

W#$8/*-"
D*()

^$-/*-

N*&/2"^$-/*-

D&,8',0"](#)$-8

N*&/2"T(##*5

T(-%$#"3(-$

Y,''$8)$-"C&-+/,*-

N*&/2"
T(.08/$()

^,'E&#-"
T,%2"D*()

_(',-%"
V*..*-

T$(/2#*5"
[$#.,-('"c

T*&-8'*5" T*&-8'*5"
_(8/
O8/$#'$R
Y$8/

N*&/2"
_(',-%

T(//*-" T*&-8'*5" S*8/*-" Z*#/21,$')8

V#*88
A(-*#
V$-/#('
T$(/2#*5"
[$#.,-('"H

W()),-%/*S(R85(/$#
a&$$-85(R

T,%2"N/#$$/"
^$-8,-%/*-

^$-8,-%/*-"
=O'R.0,(>

N*&/2"
!+/*-

V2,85,+Q"
W(#Q

[&#-2(."
]#$$-

T(..$#8.,/2
N/(.1*#)"
S#**Q

S(#*-8"
V*&#/

TR)$"W(#Q"V*#-$#
N/?"
C(.$8`8"
W(#Q

X,+/*#,(
N'*(-$"
Nd&(#$

]'*&+$8/$#"
D*()"

Y$8/"
^$-8,-%/*-

D(7$-8+*&#/"
W(#Q

[*//$-2(."
V*&#/"
D*()

OB1*#)"
V,#+&8

]#$$-"W(#Q

N*&/2"
^$-8,-%/*-

W,++(),''R"
V,#+&8
V2(#,-%"
V#*88

Y$8/.,-8/$#

T*'E*#-

V$-/#('"3,-$

C&E,'$$"3,-$

X,+/*#,("3,-$

V,#+'$"3,-$

A$/#*0*',/(-"3,-$

Y(/$#'**"U"V,/R"3,-$

P,8/#,+/"3,-$

S$#.*-)8$R
3*-)*-"
S#,)%$

3(.E$/2"
Z*#/2

W&/-$R"S#,)%$
V'(02(."C&-+/,*-

[**/,-%"S$+

Y,.E'$)*-"W(#Q

M-/$#+2(-%$"
N/(/,*-8

!"#$%&'(#")*)$+(',-$(#"),(%#(..(/,+".(0"*1"/2$"3*-)*-"4-)$#%#*&-)")#(5-"5,/2"2*#,6*-/('"
(-)"7$#/,+('"',-$8"(-)"9:;"(-)"<:;"),(%*-('8"=/(''"7$#8,*->?"@"A(B5$''"C?"D*E$#/8F"GHI:JIK::L

V*'',$#8"Y**)

Y,.E'$)*-

N*&/2"Y,.E'$)*A*#)$-

V(-()("
Y(/$#
N&##$R"a&(R8

V'(02(."Z*#/2

Z$5"V#*88"](/$
T*-*#"O(Q"W(#Q

O7('
N/*+Q5$''

S#*+Q'$R
\*#$8/"T,''

V'(02(."V*..*V'(02(."N*&/2

N*&/21,$')8

[**/,-%"S#*()5(R

N/(/,*-

_'$02(-/"U"V(8/'$

S('2(.

3*-)*-"O7$#%#*&-)

Y$8/1$##R

Y(00,-%

^$--,-%/*-

Z*#/2$#-"3,-$
P*+Q'(-)8"3,%2/"D(,'5(R

N2()5$''

X(&B2(''
M.0$#,('"
Y2(#1

NR)$-2(.

S#,B/*-

S*5"
D*()

N/$0-$R"
]#$$Y2,/$+2(0$'

A*-&.$-/

S*#*&%2

_(8/"W&/-$R

W,++(),''R"3,-$

A,'$"
_-)
S$/2-('"
]#$$-

D*/2$#2,/2$

^$5"](#)$-8

!"#$%&'(&'$)'(#*+*&)'(

T(+Q-$
Y,+Q

[*5$#"
](/$5(R 3,.$2*&8$

S(-Q

_.E(-Q.$-/

\&'2(."
S#*()5(R

T(..$#8.,/2"U"V,/R"3,-$

!')%(/$"
_(8/
!')%(/$
[*5$#"T,''

N*&/25(#Q

W(#8*-8"]#$$-

N2*#$),/+2

[$.0'$

Y(/$#'**

W,.',+*

D,+2.*-)

T*.$#/*-

T(+Q-$R"
V$-/#('

3,7$#0**'"N/#$$/

V*7$-/"](#)$V(--*-"N/#$$/
3$,+$8/$#"
A(-8,*-"T*&8$
Nd&(#$
S'(+Q1#,(#8

Y$8/"S#*.0/*-

]&--$#8E&#R

S(Q$#'**"3,-$

O')"N/#$$/

V2(-+$#R"
3(-$
N/"W(&'`8
S*-)"
N/#$$/

P('8/*-"
^,-%8'(-)

T*B/*-

\(##,-%)*S(#E,+(A**#%(/$
D&88$''"
Nd&(#$

]**)%$"
N/#$$/

^-,%2/8E#,)%$
_(#'`8"
V*&#/

_&8/*-"
Nd&(#$

A(#E'$"
!#+2

3(-+(8/$#"
](/$

]*')2(5Q"D*()

!+/*-"
[*5-

Y(##$-"
N/#$$/

_)%5(#$"
D*()

Z*//,-%"T,''"
](/$

N2$02$#)`8"
S&82"

D$%$-/`8"
W(#Q

N/#(/1*
M-/$#-(/,*

T(%%$#8/*!-%$'

S(Q$#"
N/#$$/

W()),-%/*-

T*''(-)"
W(#Q"

Y2,/$"
V,/R

^,-%`8"V#*88"
N/?"W(-+#(8

_&8/*-

A(#R'$E*-$
_)%5(#$"D*()

A(,)("X('$

Y('/2(.
a&$$-

P('8/*-"C&-+/,*-

S(#-8E&#R

]#$(/"W*#/'(-)"
N/#$$/

Disadvantages
Y(#5,+Q"
3()E#*Q$"]#*7$
!7$-&$
Very difficult to implement well, needs practice. Poor designs can
Y$8/E*&#-$" D*R('"
W(#Q O(Q
lack visual coherence. Users may resist. Very poor designs can shock. 3(/,.$#"D*()
_(',-%"

N*&/2"
[*//$-2(.

A(-*#"T*&8$

\,-8E&#R"W(#Q

N/?"C*2-`8"Y**)

^,'E&#-"W(#Q

S#*()5(R
Z*#/2"
_(8/"
Y$8/"
Y**)"3(-$
Examples
!+/*!+/*!+/*Famously used to good effect on Moscow Metro maps in the!+/*-"
1970sN2$02$#)`8"
V$-/#('
S&82"A(#Q$/
(below). Has been tried from time to time in Paris.

N$7$-" [*//$-2(." S'(+Q2*#8$" Y('/

D*()
N,8/$#8
T('$
V

T(##,-%(R"
]#$$-"3(-$8

[&1-$''"W(#Q
^$-/,82"
[*5-"
Y$8/

V#$8+$-/

^$-8('"
]#$$a&$$-`8"W(#Q

Z*#/2"
_(',-%

V#*&+2"
T,''

S$'8,6$"W(#Q
^$-/,82"[*5Comment ^,'E&#\,-+2'$R"
D*()"U"
!#8$-('
\#*%-('
Should only be attempted
by
a
designer
who
has clear objectives
that
V2('Q"\(#.
V(.)$-"
Y$8/"T(.08/$()
S#*-)$8E&#R
D*()
T*''*5(R"D*()
cannot
be
satisfied
by
using
lower
linearity.
For
example,
S#*-)$8E&#R"
V(.)$-"
W(#Q
\,-+2'$R"D*()
T,%2E&#R"
[*5V('$)*-,(-"D*()
considerably
reduced
numbers of corners
in Central
London
areV(-*-E&#R
U"M8',-%/*^$-8('"D,8$ (below). N5,88"V*//(%$
possible
V('$)*-,(-"
A*#-,-%/*-"
D*()"U"

T(#'$8)$-

W(#Q"D*R('

[&#-0,Q$"3(-$
!#+25(R
400$#"T*''*5(R

]*80$'"
O(Q

Future Directions for the Washington, DC Metro Map

Y,''$8)$-"]#$$-

Y$.E'$R"V$-/#('
B.4 Higher-Order Linearity (e.g. five angles: decalinear;
six angles:
N&)E&#R"T,''
dodecalinear, etc.)
N/*-$E#,)%$"W(#Q
Z*#/2*'/

Advantages
]#$$-1*#)
A skilled designer can reduce the number of corners on a map
!'0$#/*W$#,7('$
without sacrificing topographical accuracy.
Best suited for very
complicated networks or strongly radial networks.

T(.08/$()"
T$(/2

P*'',8"T,''

N*&/2"D&,8',0

N&)E&#R"[*5-

T(.08/$()

Z$(8)$-

Z*#/2"Y$.E'$R

T$(/2#*5"
[$#.,-('8"
"GF"KF"9

]*')$#8"]#$$-

Y$.E'$R"W(#Q

M+Q$-2(.

Y**)"]#$$T,%2%(/$

V#R8/('"W('(+$
W$-%$"Y$8/
!-$#'$R
Z*#5**)"C&-+/,*Y$8/"V#*R)*-

Page 20

Z$5"V#*88

Future Directions for the Washington, DC Metro Map

B.5 Curvilinear (No straight lines permitted)

Advantages
Can tame the most complex networks where the only alternative is a
maze of zigzags. Excessive topographical distortion can be avoided.
For around 40% of users, a well-implemented design is flowing,
organic and engaging.
Disadvantages
Open to abuse by poor designers. Needs care to implement well and
there are no established aesthetic standards. Resisted by those users
who prefer regular geometry.
Examples
Not currently adopted, but designs have
been produced with significant usability
benefits compared with official maps in
Madrid and Paris (right) [2,3].
Comment
Although this style shows the most benefits
for complex networks, pleasing designs can
result for simple ones too. The intention of
the designer should be to replace the
zigzags on a conventional map with almostimperceptible changes in curvature as the
eye follows the lines.

Page 21

Future Directions for the Washington, DC Metro Map

Key bibliography
[1] Ovenden, M. (2007). Transit maps of the world. New York:
Penguin.
[2] Roberts, M. J. (2010). Henry Beck rules, not OK? Breaking the
Rules of Diagrammatic Map design.
www.tubemapcentral.com/Breaking_the_rules.pdf
[3] Roberts, M. J., Newton, E. J., & Lagattolla, F. D. (2010). Objective
versus Subjective Measures of Metro Map Usability:
Investigating the Benefits of Breaking Design Rules.
www.tubemapcentral.com/Roberts_Metro.pdf
[4] Schrag, Z. M. (2006). The great society subway. A history of the
Washington Metro. Baltimore: Johns Hopkins University Press.

Page 22