The Dynamics and Diversity of Space Use in the British Library

Home , British Library, Colin St John Wilson

ITU A|Z • Vol 12 No 3 • November 2015 • 23-39 Te dynamics and diversity of space use in the British Library

Kerstin SAILER [email protected] • Space Syntax Laboratory, Te Bartlett School of Architecture, University College London, 140 Hampstead Road, London NW1 2BX, United Kingdom

Received: September 2015 Final Acceptance: October 2015 Abstract Te Space Syntax study of buildings typically distinguishes between weak and strong programming, where social behaviours either follow or defy the spatial logic of a building. Tis is ofen based on analysing collective and aggregate patterns of behaviour. Tis paper builds on recent work redefning our understanding of weak and strong programming, yet aims to analyse usage patterns and spatial afordances in a much more fne-grained way by taking diversity of user groups as well as the temporal unfolding of behaviours into account. Te British Library acts as a case study and is investigated based on a rich empirical dataset of observed user behaviours. Results suggest that the British Library shows both strong and weak programming: movement fows only partially followed spatial confguration, and the interface the building constructed kept people apart rather than bringing them together. In addition, large variations in user activities existed in some parts of the Library, all of which points towards strong programming. At the same time however, certain activities showed clear spatial preferences and signifcant diferences in local and global visibility patterns, which illustrates weak programming. It was also shown how dynamic and diverse user behaviours emerged in the British Li- brary, highlighting the need to draw a nuanced picture of usage. Te contribution of the paper thus lies in a detailed and deep analysis of usage patterns, unpacking variations in behaviours between diferent users at diferent times and linking this both to the afordances of confguration as well as programmatic infuences.

Keywords Public libraries, Space syntax, Space usage, Temporal dynamics, Strong and weak programming. 24

1. Introduction: More than just If we consider buildings based on books… usage, change becomes essential. Pub- Buildings are dynamic settings lic libraries, like most other building that accommodate a range of difer- types have seen a dramatic change in ent uses. A hospital for instance is how they are used, perceived and ex- not just a place for curing the ill, but perienced. Te increasing digitisation also a workplace for nurses, doctors, of content means that new ways of ac- cleaners and porters (see for instance cessing collections emerge, thus shif- Heo, Choudhary, Bafna, Hendrich, ing the necessity of a physical site away & Chow, 2009 on nurses assignments from providing access and towards and behavioural patterns). A school other uses. Drawing on a study of 24 not only enables teaching and learning, recently built monumental public li- but is also an important place for so- brary buildings, Shoham and Yablonka cialisation, making friends and hang- (2008) came to the conclusion that the ing out (Minuchin & Shapiro, 1983; new-built libraries had increased user Sailer, 2015). Shops have clear social numbers, were full of life and served functions beyond their retail purpose wider purposes as symbols of culture, (Koch, 2014). Museums do not only as tourist attractions, but also as pleas- exhibit pieces of art, but also educate, ant meeting places in a quiet cultured entertain and sell merchandise (Kwon environment. & Sailer, 2015). Likewise, libraries have Te British Library forms a partic- always been meeting places for com- ularly interesting case in this context. munities in addition to storing books Te architect of the British Library, and organising access to information Sir Colin St John Wilson (1998) de- (Capille & Psarra, 2015). scribed the multitude of functions to All of the above descriptions cen- be accommodated as: a day-to-day tre on what people do in buildings. workplace, an institution that embod- Indeed, most buildings are defned by ies and celebrates national memory, a functions or so called ‘use types’ (For- storage of collections, places of study, ty, 2000) or ‘activity types’ (Steadman, exhibitions of its treasures, an event- 2014): a hospital is a hospital because space hosting lectures and seminars, of what happens there, and likewise and back-of-house functions such as a school is a school again because of conservation laboratories and admin- what goes on inside it (Hillier, Hanson, istration. Tis already points to a real & Peponis, 1984). diversity of space usage patterns. How Yet, it seems that what people do the publicly accessible areas of the Li- in buildings becomes even more im- brary are indeed used in their every- portant, as a new focus on the human day functioning will be explored in side of architecture as well as on peo- this paper, drawing on a rich data set ple’s experiences, behaviours and us- of empirical and both quantitative and age patterns can be observed in recent qualitative participant observations, discourses. Te question of usage and collected in 2009 and 2010 by MSc daily life has already been popular in students at the Bartlett, UCL. It will be the 1970’s with architects like Herman asked how people move around in the Hertzberger defning architecture as building, to which degree the spatial concerned with ‘daily life lived by all layout (analysed with Space Syntax) people’ (Hertzberger, 1991), however informs usage patterns and how usage only recently, scholars have argued that varies between diferent user groups, the social agenda of architecture has but also over time. Its main aim is to too long been a blind spot that needs provide a sketch of the multi-function- re-addressing (Cupers, 2013). Other ality of the building and describe user recent publications on usage and the groups and usage patterns in as much social role of architecture and design detail and variation as possible. Tis is (Awan, Schneider, & Till, 2011; Berg- an important task, if we want to refect doll, 2010; Maudlin & Vellinga, 2014; on how to design ‘social’ buildings in Till, 2009) underlined the import- the future, where usage and people’s ant reading of buildings as ‘lived in’ activities, preferences and experienc- (Brand, 1994; Hollis, 2009). es are actively anticipated, embedded,

ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 25 and allowed to grow and change. of a strong programme building, since Tis paper is structured as follows: diferent user groups with varying de- Chapter 2 will provide theoretical grees of inhabitant or visitor status foundations for the empirical explora- such as judges, barristers, witnesses, tions of usage patterns in the British Li- defendants and public were channelled brary by sketching research on library through the building along separat- buildings and human behaviours, but ed paths so that their movement was also by elaborating on the Space Syn- highly controlled and encounters were tax theory of ‘strong and weak pro- actively hindered until all users met gramming’ in buildings. Chapter 3 in the highly orchestrated and ritual- will introduce the British Library as a ised court room proceedings (Hanson, case study, followed by a detailed de- 1996). In contrast, buildings were seen scription of methodology in Chapter as weakly programmed if the interface 4. In four consecutive steps, Chapter 5 between user groups was not con- will discuss the diversity and dynam- trolled and everyone could encounter ics of usage patterns in the British Li- everyone else freely, following ‘short brary and a fnal Chapter 6 will draw models’ with a high degree of randomi- conclusions, discuss limitations of the sation and morphogenesis (Hillier & study and provide an outline of future Hanson, 1984). Te most used example research in the feld. for traditionally weakly programmed buildings was the editorial foor of a 2. Usage patterns and building types: newspaper, which fourished through On strong and weak programming generative and unstructured encoun- In their paper ‘Visible Colleges’ Hill- ters among diferent users. ier and Penn (1991) conceptualised Te implications of this theory for buildings as either strongly or weakly the understanding of buildings and programmed depending on the de- usage patterns lie mainly in the ques- gree to which the activity patterns in- tion how closely movement fows and side the buildings followed strict rules, resulting patterns of encounter corre- procedures and models. Tis theory is spond to spatial confguration. Tradi- crucial in understanding the relation- tional Space Syntax theory would sug- ship between spatial layout and usage gest that movement fows are highest patterns inside diferent building types. in areas of high spatial integration – so A programme was defned as “not called ‘natural movement’ (Hillier & the organisation it houses (…) [but] the Iida, 2005; Hillier, Penn, Hanson, Gra- spatial dimension of an organisation, jewski, & Xu, 1993), however, adding and the key element in any programme strong and weak programming, we is the interface, or interfaces, that the would only expect this relationship to building exists to construct (…) [i.e.] the hold in the case of weak programming, spatial relation between or among two where randomisation is at play and broad categories of persons (…) that the layout can act morphogenetical- every building defnes: inhabitants, or ly. In contrast, it could be argued that those whose social identity as individu- movement fows follow programme in als is embedded in the spatial layout and strongly programmed buildings. who therefore have some degree of con- Over recent years, the theory of trol of space; and visitors, who lack con- strong and weak programming was trol, whose identities in the building are taken up by diferent researchers and collective, usually temporary and sub- articulated further, for instance Koch ordinated to those of the inhabitants”. and Steen (2012) proposed a new cri- (Hillier & Penn, 1991, p. 33) terion for strong programming, thus Terefore, buildings were consid- adding more nuance and variation to ered strongly programmed if the inter- the original concept. Likewise, Cap- face between user groups was highly ille and Psarra (2013) suggested that controlled and the patterns of encoun- the unequal distribution of activities ter followed so called ‘long models’ across diferent spaces and functional with a high degree of prescription areas of a building meant strong pro- and determinism (Hillier & Hanson, gramming, whereas an equal distribu- 1984). A court was the classic example tion highlighted weak programming.

Te dynamics and diversity of space use in the British Library 26

Studying two public libraries in Lon- ferent forms of knowledge representa- don, it was concluded that one library tion (tree-like, network-like, as a con- was weakening the infuence of pro- trol system) were found. Overall, it was gramme on activities, while the other concluded that contemporary libraries one was strengthening it. could be seen as systems that increas- Additionally, it was shown that el- ingly aim to integrate people and pro- ements of strong programming could mote social encounter rather than keep appear in traditionally considered people apart by providing silence, soli- weakly programmed building types tude and concentration. To that end, it such as workplaces and ofces, for in- was shown that the activity of reading stance in the form of attractors that occurred mainly next to heavily used may defect movement fows away corridors and areas of movement fows, from spatially integrated areas (Sailer, thus giving rise to social encounters. 2007, 2010). Similarly, buildings con- Tis very phenomenon of reading sidered strongly programmed such as in close proximity to highly integrat- hospitals could show aspects of strong ed areas was found in a study of nine and weak programming, even within a academic libraries in Portugal as well, single case depending on which criteri- however, here it was reported as a on was applied (Sailer et al., 2013). noise problem inhibiting concentra- What can be learnt from these stud- tion (Both, Heitor, & Medeiros, 2013). ies, is the insight that space usage ac- Another recent study on two academic tivities are dynamically enfolding sys- libraries in London (Zong, 2015) fo- tems, embedded in spatial situations cused on the diversity of activities as a and practices, driven by organisation- result of new pedagogic ideas and dig- ally defned roles and programmes, but ital access. Analysing both spatial con- also distributed in space by confgura- fguration and furniture arrangements tion. Building types (such as hospitals, as afordances for usage, it was pro- libraries or ofces) cannot be associat- posed that a diversity of spatial charac- ed with one type of programming per teristics allowed for a diversity of usage se; neither does a particular building patterns to unfold. necessarily follow clear categorisations Te theme of libraries changing to as strong or weak programme. Te the- accommodate diferent functions was ory of strong and weak programming also the subject of a syntactic study of of buildings can help scrutinise phe- 18 public libraries in France, where it nomena, yet detailed analysis is needed was investigated how traditional librar- before a judgement on the degrees and ies with closed collections and a central levels of programming in its interplay catalogue changed into so called ‘medi- with spatial confguration can be made. atheques’ providing access to a diver- For the study of libraries, which in sity of media sources and information the traditional dichotomised descrip- types, which meant a spatial change tion of either ‘strong’ or ‘weak’ would towards open bookshelves and reading have been seen as a typical example spaces (Lim & Kim, 2009). Results sug- of a weak programme (Zook & Baf- gested that newer building types which na, 2012), this means scrutinising the followed the mediatheque model had space for aspects of strong program- lower overall values of visibility on ming (rules, procedures, attractors, average, but also a wider and more di- strong distribution of activities by verse range of confgurational options. function, movement fowing against Visibility relations also feature in confgurational logic) in addition to the paper by Zook and Bafna (2012), understanding weak programming which highlights how everyday activ- and spatial practices. ities (borrowing a book, attending a Particular studies on libraries within scheduled meeting, meeting a friend in the framework of Space Syntax seem the reading rooms) in the Seattle Pub- worth mentioning, too. lic Library – a building with a highly With the aim to discover how spa- unusual spatial composition – still tial systems produce meaning, Koch follow genotypical patterns of visual (2004) studied three public libraries in access, where paths lead through ex- Sweden and concluded that three dif- pected levels of openness and enclo-

ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 27 sure, thus publicness and privacy, un- tings staging a multitude of diferent derlining the view of the library as a usage patterns beyond the traditional known institution. Tis so called ‘social access to collections and the accom- staging’ is contrasted by a view of the modation of the process of reading. ‘phenomenal staging’ – the subjective, Instead libraries were shown to be so- individual experience of a user as they cial spaces, experienced diferently by make their way through the building, people and supported by specifc con- which is characterised by unexpected fgurational properties of the library vistas, hidden outlooks and surprising buildings. elements. How these phenomena resonate in Libraries were also used as settings the case of the British Library will be to understand issues of wayfnding and explored in the following chapters. signage (Carlson, Hölscher, Shipley, & Conroy Dalton, 2010; Li & Klippel, 3. Case study: Te British Library 2010, 2012) and to test new methods Tis paper draws on rich observa- for user feedback and Post-Occu- tions of space usage in one particu- pancy Evaluations (Dalton, Kuliga, & larly interesting building: the British Hölscher, 2013). Library. As National Library of the In summary, previous research has United Kingdom, its aim is to store shown how libraries have become set- every book published in the UK and

Figure 1. Annotated foor plan of the British Library.

Te dynamics and diversity of space use in the British Library 28 make its collections freely accessible to 2004, p. 79), able to adapt to future 1Some of the books the public. With a building size of more usage and organisational needs, for of the British Library are stored than 122,000 sqm it stores around 170 instance the new ways in which infor- in its branch in million items (among them almost 14 mation will increasingly be stored and Boston Spa in million books)1 and employs around accessed digitally. Yorkshire, from 1200 staf in its main St Pancras build- where they can ing. As an institution, the British Li- 4. Methodology be ordered to the Reading Rooms brary was founded in 1972 by an Act Tis paper combines the syntac- in St Pancras of Parliament, but frst continued hav- tic study of the British Library, based within 48 hours; ing its collection and reading rooms on axial accessibility maps (drawn on basic statistics housed in the British Museum, until it knee level) and Visibility Graph Anal- are from Wilson moved to its own site in 1997, when the ysis (VGA, constructed on eye level) 1998 and updated statistics on the frst reading room opened. (Turner, Doxa, O’Sullivan, & Penn, building are taken Te British Library’s main building 2001) with detailed and structured ob- from: https:// in the centre of London near the rail- servations of space usage patterns. Te en.wikipedia. way station St Pancras was fully opened following three standard Space Syn- org/wiki/British_ to the public in 1998 afer a more than tax observation techniques (Al-Sayed, Library thirty-fve year-long conception, de- Turner, Hillier, Iida, & Penn, 2015; sign and construction process. Te Grajewski, 1992) were used: gate- British architect Colin St John Wilson counts, traces by following people and created a building ‘based on purpose’ snapshots. and human scale in an approach that For the gate-counts, movement he called ‘the other tradition’ of Mod- fows across a total of 127 imaginary ernism (Wilson, 2007). For Wilson, ar- gates on all six public foors of the chitecture was grounded in use, creat- British Library were counted for fve ing an ordered framework for activities minutes each in the morning, midday to happen; this was also called an ‘ar- and afernoon on three days (including chitecture of experience’ (Stonehouse, Saturday) in 2009 and on two days (in- 2004). Te building was described as cluding Sunday) in 2010. Gender and 2 ‘inviting’ and ‘democratic’ by critics: readership status was recorded. Data 2 “A library might be expected to be was aggregated across all observations Building users conceived as a monument. (…) Yet were distinguished and collective hourly fow at each gate by visual cues: this building [the British Library] has was calculated. Readers could found a kind of democratic equiva- Movement was also captured be recognised by lent. (…) Tis kind of monumentality through traces, where observers picked their Reader’s is not imposed upon us; it is assigned up building users at entrances or other passes, sometimes by us. So the building is symbolic, but worn around this symbolism is not assertive and it is movement distributors on each foor people’s necks, or not about great occasions or collective plate (lif, staircases) and discretely fol- more obviously, events. Te building seeks relationships lowed them for 10 minutes (in 2009), by carrying their with the individuals who use it and vis- 5 minutes (in 2010) or until they had belongings in it it, through a sense of invitation frst plastic carrier bags, reached a destination (for instance a which were the evident in the forecourt and entrance. desk in a Reading Room or a seat in only bags allowed You are invited to be a participant, the café), or in fact lef the foor plate inside the reading not merely a spectator.” (MacCormac, or building. Te route they took was rooms. Tis means 2004, pp. xii-xiii) traced on a foor plan and digitised that everyone Te way in which the building ne- in GIS. A total of 679 building users categorised as gotiates between individual usage and a Reader in the were shadowed during Library open- observations institutional representation was de- ing hours, and additional demographic defnitely had a scribed as ‘intimate monumentality’: and user specifc information (gender, Reader’s pass, but “Te building is book-like, reveal- estimated age range, formal or infor- the observations ing its inner world only when entered, mal attire, Reader or Non-Reader) was could be biased an individual, intimate act. Te hard, by not recognising rather sober exterior allows the inte- noted. all Readers as rior to be revealed and discovered on Snapshots recorded the exact loca- such (for instance entering and using the library (…) – all tion and type of activity of building us- those leaving their part of an intimate experience of mon- ers at a precise moment in time. All six possessions in umentality.” (Stonehouse, 2004, p. 69) lockers or at a desk publicly accessible foors of the build- in the Reading It was also praised to provide an “in- ing were observed repeatedly through- Rooms and going herent versatility of form” (Stonehouse, out the course of the day. Most areas for a cofee).

ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 29 were observed eight times in total with gle relevant correlation was found, al- more intensively used areas captured though Choice and Global Integration up to 17 times to get a higher data res- yielded highly signifcant / signifcant olution. Primary and secondary activ- results for Non-Readers (p<0.0068** ities were recorded: primary activities and p<0.0142* respectively) yet with included sitting, standing and walking very low R2 values of 0.03 each. Clear- as individual behaviours and interac- ly, spatial confguration cannot explain tions as group behaviours, while sec- the overall distribution of moving peo- ondary activities distinguished a total ple very well in this case. Interestingly of 22 diferent behaviours (such as better correlations appear for a foor by searching, shopping, working on a lap- foor analysis, particularly for the up- top, reading, looking at an exhibition) per foors and for the non-reader de- and combinations of behaviours (such mographic, as shown in Table 1. as laptop and reading, looking at an It seems that building users do fol- exhibition and talking, etc.). A total of low the confgurational logic of space 7993 people were observed. to some degree when they are moving through the building, yet this is only 5. Te British Library: Diversity and the case for Non-Readers and only for dynamics of usage patterns the 1st, 2nd and 3rd foors, where correla- Patterns of usage in the British Li- tion coefcients higher than 0.20 were brary will be analysed in this section, achieved. Even then this is not a very discussing the distribution of activities strong relationship. in space, the diversity of behaviours, Various factors interfere with the rhythms and temporal patterns as well confgurational logic. It could be ar- as emerging communities and their gued that Readers know the building specifc needs and activities. well and come for a particular purpose, i.e. to use the collections and take 5.1. Movement fows a seat in one of the Reading Rooms. As a frst step in the analysis, it is of Terefore, their movements are much interest to understand the overall dis- more programmed and as such do tribution of people in the building and not follow confguration. Secondly, investigate to which degree movement the ground foor as well as the foors fows are driven by confguration (in- above (mezzanine) and below (lower dicating a weakly programmed build- ground) provide many diferent facili- ing) or in contrast by programme and ties and places of interest, most of them function (indicating a strongly pro- specifcally targeted at Non-Readers, grammed building). such as the shop, exhibition spaces, Four spatial variables of the axial the café and canteen, the information map were analysed regarding their re- desk, the cloak room, etc. Tis means lationship to the overall fow of people attractors (Sailer, 2007) divert the fow in the British Library, as well as to the of movement of Non-Readers and may fow of Readers and Non-Readers (us- counteract confguration as a way to ing gate-count data): Connectivity, In- distribute people. Te role of the en- tegration Radius 3 (Local), Integration trance should not be underestimated Radius N (Global) and Choice. No sin- either. Every single building user pass-

Table 1. Coeficient R2 for correlation of movement fows (total, Reader, Non-Reader) with global Integration [INT] and Choice [CHOI].

TOTAL READERS NON-READERS Floor R2 [INT] R2 [CHOI] R2 [INT] R2 [CHOI] R2 [INT] R2 [CHOI] Whole Building 0.00 0.01** 0.00 0.00 0.03* 0.03**

Lower Ground -0.01 -0.02 -0.07 -0.02 -0.00 -0.00 Ground 0.08 0.01 0.08 -0.03 0.00 -0.00 Mezzanine 0.12* 0.04 0.05 0.03** 0.15** 0.04** 1st Floor 0.14** 0.20** 0.00 0.04 0.25** 0.16** 2nd Floor 0.23** 0.28** 0.12 0.00 0.20** 0.07 3rd Floor 0.05 0.34** 0.14 -0.01 0.29** 0.24** * Values marked in bold with * were significant at the 0.05 level and ** at the 0.01 level. Negative correlations are shown in green, low R2 (<0.2) are shown in grey and above that in black.

T e dynamics and diversity of space use in the British Library 30

Figure 2 a-d. Movement traces of 427 Readers (red) and 149 Non-Readers (black) on the ground foor, mezzanine, first and second foor. es through the entrance, however it is as the ground foor, mezzanine and frst not necessarily also the most integrat- foor, yet, there are many spaces with a ed place in the building. Buildings with distinctive dominance of either Read- many foors ofen show the integration ers (upper foors, circulation, staircas- core placed around the geometric cen- es, Reading Rooms) or Non-Readers tre of the building as a whole, which is (exhibitions, café, canteen). It can also also is the case for the British Library, be seen from the traces that Readers where the most integrated areas are (shown in red in Figure 2a-d) move in found on the frst foor. Again, this a rather targeted fashion with straight distorts the correlation between move- routes, while Non-Readers (shown in ment fows and confguration. black) tend to wander more aimlessly Te relationship between difer- along curvy paths. ent user groups such as Readers and Te degree of co-presence between Non-Readers can also be analysed as a Readers and Non-Readers can also be matter of the interface constructed by investigated statistically by correlating the building to bring people together or total numbers for each group across keep them apart. First of all movement the diferent locations in the building. traces of the two diferent user groups With gate-count data a correlation of can be compared visually. It can be seen R2=0.28, p<0.0001 is obtained, show- in fgure 2a-d that movement fows being that Readers and Non-Readers tween Readers and Non-Readers over- distributed rather diferently across lap in certain parts of the building such the building: areas with high counts

ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 31 of Readers showed rather low counts and interactions as group behaviours, of Non-Readers and vice versa. Tis highly signifcant diferences of con- means the building creates a controlled nectivity and mean depth can be found interface between the diferent user between these activities in a statistical groups and keeps them apart rather Analysis of Variance (ANOVA). Re- than bringing them systematically to- sults of the ANOVA tests are shown in gether. It seems the Library ofered at- Table 2; Figures 3a-b show the VGA of tractive spaces to each group separate- the building for connectivity and mean ly; they co-existed rather than cohered depth. Diferences in connectivity are and came together. more pronounced, leading to a higher In summary, the analysis of move- coefcient of R2=0.07 (p<0.0001), while ment fows has highlighted that the mean depth only shows an R2=0.01 British Library is a predominantly (p<0.0001), yet both are highly signif- strongly programmed building: over- cant efects also due to the large sample all fows do not follow confguration- size. Te fact that connectivity seems to al logic consistently and diferent user relate more strongly to user behaviour groups with distinct usage patterns is in line with fndings reported in pre- (Readers vs Non-Readers) were sepa- vious research (Haq, 2003). rated to a high degree. A further analy- In detail, interactions have the lowest sis of strong and weak programming in average connectivity (284 VGA pixels, buildings as a function of the diversity which equates to an area of 284 sqm, and distribution of activities will follow since the VGA grid was set to 1x1m); in the next section. interactions are also relatively high in mean depth compared to other activ- 5.2. Diversity and distribution of ac- ities (5.83 on average), which means tivities people interacted in rather segregated To investigate diversity and distribu- and smaller areas. Since the average tion of activities, a two-step approach values of connectivity and mean depth was followed: frstly it was analysed for the whole building are 585 and 5.95 whether primary and secondary activ- respectively, all observed activities ities difered according to their spatial were more integrated than the building properties of connectivity and integra- average (<5.95). Standing and walking tion (i.e. visual Mean Depth), retrieved took place in medium sized areas (354 from the VGA. Tis will highlight and 418 sqm on average), but walking whether certain activities show prefer- clearly happened in the most integrat- ences for areas with high or low direct ed places (lowest average mean depth visibility (connectivity) and for areas of 5.54), while standing occurred in with strategically short or long visual the most segregated ones of the places paths (mean depth). Secondly, the dis- observed (MD=5.89). Sitting enjoyed tribution of activities will be brought the largest view-sheds with an average together with the functional alloca- of 606 sqm, which is larger than the tions of spaces to analyse whether spe- building average; this is clearly due to cifcally allocated areas attract usage sitting being most prominent in the diferently from the overall building reading rooms, which are also relative- averages. ly large in size. Regarding the spatial logic of pri- Secondary activities also showed mary activities, i.e. sitting, standing signifcant diferences between the di- and walking as individual behaviours rect and strategic visibility of various

Table 2. Number and statistics [mean, standard error] of spatial properties connectivity [CONN] and mean depth [MD] of observed primary activities from ANOVA tests. Activity Count Mean [CONN] Std Err [CONN] Mean [MD] Std Err [MD] Interaction 591 284.836 16.535 5.83134 0.03167 Sitting 6137 606.156 5.131 5.75574 0.00983 Standing 904 354.165 13.370 5.88449 0.02560 Walking 697 418.305 15.226 5.53567 0.02916 * Connectivity values larger than the building average and mean depth values lower than the building average are highlighted in red.

Te dynamics and diversity of space use in the British Library

Figure 3 a-b. Visual Graph Analysis of the British Library: Connectivity (a) and Mean Depth (b). behaviours; the ANOVA results for an exhibition could be argued to be connectivity were highly signifcant driven by the functional programme (p<0.0001) with an R2=0.27, while rather than the preference for a par- R2=0.10 was obtained for mean depth ticular spatial character of integration (p<0.0001). Details are presented in or segregation, hence the distribution Table 3. Interestingly for instance, lap- of activities by function will be inves- top users preferred smaller and more tigated next. integrated areas, especially if they were In order to do so, the 22 diferent also talking (CONN=265, MD=5.47), observed activities3 were clubbed to- whereas laptop users reading in par- gether into ten broader core activities allel sat in larger and more segregated (as listed in fgure 4a-b), for instance areas (CONN=929, MD=6.03). People all activities involving talking (Eating looking around tended to be in very Drinking Talking, Exhibition Talking, 3 Only the data collected in 2009 integrated areas (MD=4.37); similar- Laptop Talking, Reading Talking) were was taken into ly those occupants using their phones grouped together into ‘Talking’ rather account, since were also found in integrated spaces than distinguishing by additional ac- in 2010 sitting (MD=4.70). tivities. Tis procedure also ensured in some areas Tis means people seek out specif- that results were comparable to those (specifcally in the Reading Rooms) ic types of spaces in order to go about reported by Capille and Psarra (2013), was not broken specifc activities. Spatial confguration who distinguished nine diferent activ- down further (e.g. therefore played a role in distributing ities in their study of public libraries. reading, laptop activities in space according to its de- Te method presented by Capille and usage, etc.) and gree of local and global integration and Psarra to quantify the degree of pro- this would distort the following segregation. Some activities however, gramming in a library by calculating analysis. for instance shopping or engaging with the distribution of activities for the ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer

Table 3. Number and statistics [mean, standard error] of spatial properties connectivity [CONN] and mean depth [MD] of observed secondary activities from ANOVA tests.

Activity Count Mean [CONN] Std Err [CONN] Mean [MD] Std Err [MD] Bag Cloak Locker 22 213.261 77.42 6.70176 0.15607 Eating Drinking 133 211.232 31.49 6.03696 0.06348 Eating Drinking Laptop 7 159.619 137.26 6.51515 0.27669 Eating Drinking Reading 12 203.076 104.83 5.54179 0.21132 Eating Drinking Talking 115 218.767 33.86 5.88826 0.06826 Exhibition 297 341.650 21.07 6.05512 0.04248 Exhibition Talking 33 402.942 63.22 6.17684 0.12743 Laptop 819 399.693 12.69 5.40453 0.02558 Laptop Reading 311 929.028 20.59 6.03018 0.04151 Laptop Talking 34 265.319 62.28 5.47077 0.12554 Lift 4 333.188 181.57 5.93379 0.36602 Looking 19 323.947 83.31 4.37154 0.16794 Order 74 153.375 42.21 6.14963 0.08510 Phone 47 314.777 52.97 4.69967 0.10678 Reading 420 514.097 17.72 5.75063 0.03572 Reading Talking 70 155.665 43.40 6.37389 0.08750 Searching 8 195.458 128.39 6.21016 0.25882 Shopping 66 180.619 44.70 7.03216 0.09011 Sitting 3534 752.236 6.11 5.76712 0.01231 Standing 208 430.779 25.18 5.75250 0.05076 Talking 880 275.280 12.24 5.74283 0.02468 Walking 625 425.535 14.53 5.57156 0.02928 * Connectivity values larger than the building average and mean depth values lower than the building average are highlighted in red. building as a whole and comparing it to ish Library, amounting to 40% of all the distribution in diferent functional observed activities, followed by talking areas was applied here. In addition to (20%), reading (15%) and sitting (8%). investigating functional areas (Figure While the mix of activities in the cor- 4a), the variation across the diferent ridors and café strongly resembled foors was also scrutinised (Figure 4b). those found in the entire building (see Looking at the building as a whole, fgure 4a), some more variation was it can be seen that using a laptop is the found in the foyer (where 19% of peo- most predominant activity in the Brit- ple engaged with exhibition material

Figure 4 a-b. Distribution of ten core activities (based on 2009 data) across diferent functional areas [a] and foors [b] of the building in comparison to the whole building average. Te lower ground and 2nd foor were not observed in 2009. Percentage values are rounded. Te dynamics and diversity of space use in the British Library 34 rather than just 3%) and the reading jority of fow intensity across the difer- rooms (where laptop use rose to 63%); ent spaces difered between weekday the highest level of variation however and weekend. Generally speaking, the was found on the staircases and the upper foors showed higher usage in- exhibition spaces themselves. Te fg- tensity during the week, whereas low- ures obtained for the average variation er foors showed more intensive usage in those two areas are in line with the over the weekend. ones reported by Capille and Psarra for Another interesting diference be- a strongly programmed building. Tis tween weekday and weekend usage means that some parts of the British Li- patterns can be revealed by re-doing brary enact a strong programme, dis- the analysis of variation of connectivity tributing and shaping behaviours and values of each activity (as done in sec- the mix of activities in addition to the tion 5.2 and shown in Table 3 above), efect of spatial confguration, as ar- but now executed separately for week- gued in the previous section. day versus weekend. Te same was re- Analysed foor by foor, the mix of peated for mean depth. Te ANOVA activities on the mezzanine level, as is highly signifcant for all datasets well as the 1st and 3rd foors is compara- (p<0.0001); correlation coefcients for ble to the overall building distribution; connectivity were higher for weekdays only the ground foor showed signif- (R2=0.33) than weekends (R2=0.20), cant variation, which is due to the high whereas it is the other way around for number of specialised functions on the mean depth with lower coefcients for ground foor, for instance exhibition weekdays (R2=0.14) than weekends spaces. Te role of the ground foor (R2=0.34). It should be noted that co- in helping people to orient themselves efcients generally rise by splitting the also becomes obvious in the dispropor- data by day of the week, which shows tionately high percentage of people us- that diferent patterns were evolving ing their phones (14% rather than 2%) on weekdays versus weekends. Te dif- and more than twice the percentage of ferences between average connectivity people talking (43% rather than 20%). and mean depth values between week- To summarise, the British Library day and weekend are plotted in Figure is a building combining both elements 5a-b. of weak programming (since activ- Te most pronounced diferences ities showed statistically signifcant can be observed for activities involv- preferences for integrated or segregat- ing the use of laptops and talking, ed spaces) and strong programming but also for reading, walking and sit- (since activities were distributed un- ting. People working on their laptops evenly across the functional areas of preferred smaller (CONN=234) and the building). more integrated spaces (MD=5.04) on the weekends as opposed to weekdays 5.3. Rhythms and temporal patterns (CONN=451, MD=5.51); this is even of usage more pronounced for those using their In addition to the analysis of the laptop alongside reading (CONN=236 overall diversity of activities unfolding vs CONN=944 and MD=5.31 vs in the British Library, changes in usage MD=6.03). Reading itself as well as patterns over time were also investigat- sitting showed a similar preference ed. for smaller and more integrated spac- Regarding the distribution of move- es on the weekends than on weekdays, ment, it can be seen that busy areas despite the fact that weekend observa- (with high fow intensity) during the tions were done on Saturdays, which week are not necessarily those also pop- meant the Reading Rooms (as rath- ulated to a higher degree on weekends. er large and segregated spaces) were A correlation of gate counts based on open. Talking occurred in similarly traces for the week versus the weekend sized areas on weekdays and weekends, reveals an R2=0.32 (p<0.0001), which however, on weekends talking hap- means that 32% of fow intensity on pened in much more integrated spaces the weekend can be predicted by fow (MD=5.27 rather than 5.96). Possibly intensity during the week, but the ma- it could be the case that users sought

ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 35

Figure 5 a-b. Average connectivity [a] and average mean depth [b] of activities on weekdays versus weekends (based on 2009 data only, since weekend observations were not done in 2010). more socialising opportunities on tivities across functional areas again, weekends and thus preferred to place not much variation appeared between themselves in more buzzy (hence inte- weekday and weekend, however inter- grated) areas. esting diferences can be detected over Looking at the distribution of ac- the course of the day (as illustrated in Figure 6). Concentrated work such as reading or working on a laptop peaks in the af- ternoon (from 3-4pm); eating peaked at lunchtime and in the early afernoon as expected (between 1-3pm); social activities such as talking and interactions peaked mid-morning (11-12pm), at lunchtime (1-2pm) and in the late afernoon (5-6pm); and the engagement with exhibits showed a high in early morning (10-11am) afer lunch (2-3pm) and in the afernoon (4-5pm). In summary, this analysis highlights how temporal patterns of usage evolve, creating a rhythm of activities and experiences over the course of a day with Figure 6. Variation of ten core activities plus interactions between shifing preferences and locations of people (based on 2009 data) over the course of the day. Te time activities between weekday and week- marked in the diagram highlights the starting time of the observation end. Qualitative accounts of people’s period. engagement and behaviours in the

Te dynamics and diversity of space use in the British Library 36

British Library (Tomas, 2013) seem to underline this interpretation of user experience of rhythms and temporal patterns, where people chose to change activities as the day went on.

5.4. Emerging communities and usage patterns Last but not least, specifc usage patterns of emerging communities and particular groups of people can be investigated. From qualitative observations we know that a group of people using the British Library for working purposes queue outside the building every morning to take up seats in front of the Kings Library on the frst foor, Figure 7. Community of nomadic workers in front of the Kings which ofered good seating, nice indi- Library. Photograph by Kerstin Sailer. vidual lighting and power plugs in addition to the free wif available in the Tis account of emerging commu- whole building. Tose spaces shown in nities and diferences in the distribu- Figure 7 ofer both opportunities for tion of people with certain user demo- socialising (as they are in a highly fre- graphics highlights how the building quented route) as well as solitude (by afords behaviours by particular groups the nature of the furniture) and were of people in distinct ways. the most popular seats in the Library, essentially being occupied frst thing in 6. Conclusions: On diversity, dynam- the morning and throughout the whole ics and built form day. As an emerging community of so Tis paper presented evidence from called ‘nomadic workers’, people have observations of space usage patterns in come to know each other and watched the British Library in conjunction with out for other people’s belongings. an analysis of the spatial confguration Other specifc communities of peo- of the building and its afordances for ple with particular space usage patterns user behaviours. It was shown how were entrepreneurs using the IP and movement fows in the British Library Business Centre of the British Library. mostly defed confgurational logic. All areas connected with the IP and In addition to a rather controlled in- Business Centre showed dispropor- terface between diferent categories of tionate numbers of males (four times people such as Readers and Non-Read- as many males as females as opposed ers, this drew a picture of a strongly to a ratio of 1:1.15 for the building as programmed building. However, the a whole), but also higher numbers of analysis of the distribution of activities users in the age range 40-60 (1.7 older across space highlighted that activi- people per younger people in contrast ties and behaviours of people followed to a 1:1.03 ratio for the entire building). confgurationally defned preferences Other areas with an uneven distri- and thus showed weak programming. bution of users by additional demo- Functional areas in contrast, in partic- graphic information include a female ular exhibition spaces and to a smaller dominance in the Social Sciences degree the foyer and Reading Rooms Reading Room (3.4 females per male) maintained elements of strong pro- and the shop (1.6 females per male) gramming, since the mix of activities and a higher presence of older people there difered signifcantly from the (40-60 years of age) in the Philatelic overall building average, pinpointing Exhibition (3.6 older people per young the many ways in which the functional person), whereas twice as many 20-40 allocation and afordances of diferent year olds as compared to 40-60 year spaces drove usage behaviours. Te olds were found around the areas of the analysis of temporal dynamics, emerg- Folio Society Gallery. ing communities and diferent user experiences over the course of the day ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 37 and between weekday and weekend ‘builds relationships with individuals’ highlighted how the building constant- as evident in the diverse and dynamic ly evolved, shifed and changed, de- usage patterns showcased in this paper. pending on perspective. Te main contribution of this paper Acknowledgements is therefore the conceptualisation of I would like to express my sincere a building as a layered, dynamic and gratitude to the staf at the British Li- changing experience rather than as brary, in particular their Welcome a defnite entity impacting collective Team and Mark Walton, who have user behaviour in one particular way. It made it possible to investigate the also shifs the attention of Space Syntax building and its usage. I would also like analysis away from top-level collective to thank two generations of students user behaviours to more nuanced and of the MSc ‘Advanced Architectural detailed understandings of the diver- Studies’ (now called: ‘Spatial Design: sity and dynamics of the relation be- Architecture and Cities’), who have tween confguration and space usage. collected the data used in this paper Due to the nature of the used data during long days in the feld as part of set, this paper has clear limitations; is- their coursework; these are in alpha- sues include inconsistent data (for in- betical order: Eleni Alexiou, Min Hi stance diferent observation standards Chun, Monica Datta, Jean-Francois in 2009 and 2010), missing data (for in- Goyette, Birce Eren Karafazli, Eun Hye stance not all areas were covered equal- Kim, Tim Mason, Fiona McDonald, ly well), and possibly limited quality of Gillian McNally, Nikolina Nikolova, the data due to issues with interobserv- Stella Parpa, Amanda Pluviano, Ro- er reliability, specifcally given that the samund Pomeroy, Zhen (Alex) Qian, data was collected by Master’s students Carolina Rodriguez, Aabid Raheem, in their frst weeks of their degree. Khondoker Mobinur Rahman, Fer- Wherever possible those limitations nanda Lima Sakr, Ria George, Frederik were taken into account for the difer- Weissenborn and Jingcao Zhang (MSc ent types of analysis. AAS cohort of 2009-2010); and Rahwa To conclude, this paper has inves- Ayob, John Bingham Hall, Michelle tigated the diversity of diferent space Chan, Vasiliki Gogou, Rosie Haslem, usages of a building over time in rela- Zahra Khaniki, Efstathia Kostopoulou, tion to its spatial confguration. It has Annita Miltiadis, Krisangella Cama- explored both temporal dynamics as cho, Rosica Pachilova, Fei Que, Zheng well as usage diversity to incorporate a Xie and Kelin Yue (MSc AAS cohort of more diferentiated perspective on who 2010-2011). Last but not least, I would uses a building when for what purpose, like to thank Dr Jaehong Lee for his or in short the ‘multiplicities of occu- help with the spatial analysis as well as pation’ (Groák, 1992). Space Syntax with cleaning and processing observa- can ofer a fruitful framework for this tion data. exploration beyond mere aggregate and collective social patterns. Future References research could focus on the nuances of Al-Sayed, K., Turner, A., Hillier, B., temporal and user-specifc dynamics Iida, S., & Penn, A. (2015). Space Syn- more systematically to address what tax Methodology. A teaching textbook Brand (1994) called a ‘shocking lack for the MSc Spatial Design: Architec- of data’ on building usage. He high- ture and Cities Retrieved from http:// lighted the need for studies of all kinds discovery.ucl.ac.uk/1415080/ of buildings in use and what changes Awan, N., Schneider, T., & Till, J. from hour to hour, day to day, week (2011). Spatial Agency. Other ways of to week, month to month and over the doing architecture. Abingdon / New years. 20 years later this is still an open York: Routledge. research question, which this paper Bergdoll, B. (2010). Introduction. In hopes to contribute towards. A. Lepik (Ed.), Small Scale, Big Change: Trough its architecture and ‘ver- New Architectures of Social Engage- satility of form’, the British Library ment (pp. 7-11). New York: Museum of has clearly managed to be a space that Modern Art.

Te dynamics and diversity of space use in the British Library 38

Both, K., Heitor, T., & Medeiros, V. Hillier, B., & Hanson, J. (1984). Te (2013). Spaces for Knowledge: Strate- social logic of space. Cambridge: Cam- gies in academic library planning and bridge University Press. design. Paper presented at the 9th In- Hillier, B., Hanson, J., & Peponis, J. ternational Space Syntax Symposium, (1984). What do we mean by building Seoul. function? In J. A. Powell, I. Cooper & S. Brand, S. (1994). How buildings Lera (Eds.), Designing for building util- learn. What happens afer they’re built. isation (pp. 61-72). London: Spon Ltd. New York / London: Penguin Books. Hillier, B., & Iida, S. (2005). Network Capille, C., & Psarra, S. (2013). efects and psychological efects: a theory Space and planned informality: Strong of urban movement. Paper presented at and weak programme categorisation the Proceedings of the 5th Internation- in public learning environments. Pa- al Space Syntax Symposium, TU Delf. per presented at the 9th International Hillier, B., & Penn, A. (1991). Visible Space Syntax Symposium, Seoul. Colleges: Structure and Randomness Capille, C., & Psarra, S. (2015). Dis- in the Place of Discovery. Science in ciplined informality: Assembling un- Context, 4(1), 23-49. programmed spatial practices in three Hillier, B., Penn, A., Hanson, J., public libraries in Medellin. Paper pre- Grajewski, T., & Xu, J. (1993). Natu- sented at the 10th International Space ral movement: or, confguration and Syntax Symposium, London. attraction in urban pedestrian move- Carlson, L. A., Hölscher, C., Shipley, T. ment. Environment and Planning B: F., & Conroy Dalton, R. (2010). Getting Planning and Design, 20, 29-66. Lost in Buildings. Current Directions in Hollis, E. (2009). Te Secret Lives of Psychological Science, 19(5), 284-289. Buildings. From the Parthenon to the doi: 10.1177/0963721410383243 Vegas Strip in thirteen stories. London: Forty, A. (2000). Words and Build- Portobello Books. ings. A Vocabulary of Modern Architec- Koch, D. (2004). Spatial Systems ture. London: Tames and Hudson. as Producers of Meaning - the idea of Grajewski, T. (1992). Space Syntax knowledge in three public libraries. Observation Manual (2001 unpub- Licentiate Tesis, KTH Stockholm, lished revised edition: L. Vaughan) Stockholm. (2001 unpublished revised edition: L. Koch, D. (2014). Changing building Vaughan ed., pp. 18). London: UCL typologies: Te typological question Bartlett and Space Syntax Ltd. and the formal basis of architecture. Groák, S. (1992). Te idea of build- Te Journal of Space Syntax, 5(2), 168- ing. Tought and action in the design 189. and production of buildings. London: E Koch, D., & Steen, J. (2012). Analysis & FN Spon. of strongly programmed workplace en- Hanson, J. (1996). Te architec- vironments. Architectural confguration ture of justice: iconography and space and time-space properties of hospital confguration in the English law court work. Paper presented at the 8th In- building. Architectural Research Quar- ternational Space Syntax Symposium, terly, 1(4), 50-59. Santiago de Chile. Haq, S. (2003). Investigating the Kwon, S. J., & Sailer, K. (2015). See- syntax line: confgurational properties ing and being seen inside a museum and cognitive correlates. Environment and a department store. A comparison and Planning B: Planning and Design, study in visibility and co-presence pat- 30(6), 841-863. terns. Paper presented at the 10th In- Heo, Y., Choudhary, R., Bafna, S., ternational Space Syntax Symposium, Hendrich, A., & Chow, M. P. (2009). A London. modeling approch for estimating the im- Li, R., & Klippel, A. (2010). Using pact of spatial confguration on nurses’ space syntax to understand knowledge movement. Paper presented at the 7th acquisition and wayfnding in indoor International Space Syntax Sympo- environments. Paper presented at the sium, Stockholm, Sweden. 9th IEEE International Conference on Hertzberger, H. (1991). Lessons for Cognitive Informatics (ICCI), Beijing. Students in Architecture. Rotterdam: Li, R., & Klippel, A. (2012). Way- Uitgeverij 010 Publishers. fnding in Libraries: Can Problems ITU A|Z • Vol 12 No 3 • November 2015 • K. Sailer 39

Be Predicted? Journal of Map & Ge- Programmed is a Strong Programme ography Libraries, 8(1), 21-38. doi: Building? A Comparative Analysis of 10.1080/15420353.2011.622456 Outpatient Clinics in Two Hospitals. Pa- Lim, H., & Kim, S. (2009). Changes per presented at the 9th International in Spatial Organization in French Pub- Space Syntax Symposium, Seoul. lic Libraries. Journal of Asian Archi- Shoham, S., & Yablonka, I. (2008). tecture and Building Engineering, 8(2), Monumental Library Buildings in the 323-330. doi: 10.3130/jaabe.8.323 Internet Era: the future of public librar- MacCormac, R. (2004). Foreword. ies. IFLA Journal, 34(3), 266-279. doi: In R. Stonehouse & G. Stromberg 10.1177/0340035208097227 (Eds.), Te Architecture of the British Steadman, P. J. (2014). Building Library at St Pancras (pp. xii-xiv). Lon- Types and Built Forms. Kibworth Beau- don: Spon Press. champ: Matador. Maudlin, D., & Vellinga, M. (Eds.). Stonehouse, R. (2004). Composi- (2014). Consuming Architecture. On the tion and Context. In R. Stonehouse & occupation, appropriation and interpre- G. Stromberg (Eds.), Te Architecture tation of buildings. Abingdon / New of the British Library at St Pancras (pp. York: Routledge. 43-79). London: Spon Press. Minuchin, P., & Shapiro, E. K. (1983). Tomas, I. (2013). In the Library. Te school as a context for social de- London Review of Books, 35(8). Re- velopment. In E. M. Hetherington & P. trieved from http://www.lrb.co.uk/ H. Mussen (Eds.), Handbook of child v35/n08/inigo-thomas/in-the-library psychology: Socialization, personality Till, J. (2009). Architecture Depends. and social development (Vol. 4, pp. 197- Cambridge/MA: MIT Press. 274). New York: Wiley. Turner, A., Doxa, M., O’Sullivan, D., Sailer, K. (2007). Movement in work- & Penn, A. (2001). From isovists to vis- place environments - confgurational or ibility graphs: a methodology for the programmed? Paper presented at the analysis of architectural space. Envi- 6th International Space Syntax Sympo- ronment and Planning B: Planning and sium, Istanbul. Design, 28(1), 103-121. Sailer, K. (2010). Te space-organi- Wilson, C. S. J. (1998). Te Design sation relationship. On the shape of the and Construction of the British Library. relationship between spatial confgura- London: Te British Library. tion and collective organisational be- Wilson, C. S. J. (2007). Te other tra- haviours. Doctoral Dissertation PhD, dition of modern architecture: the un- Technical University of Dresden, Dres- completed project. London: Black Dog den. Retrieved from http://nbn-re- Publishing. solving.de/urn:nbn:de:bsz:14-quco- Zong, W. (2015). Higher Education sa-38427 Libraries: Exploring the Library Users’ Sailer, K. (2015). Te Spatial and Activities Facilitated with the Change Social Organisation of Teaching and of Learning Conceptions. MSc Disserta- Learning – Te Case of Hogwarts School tion, UCL, London. of Witchcraf and Wizardry. Paper pre- Zook, J. B., & Bafna, S. (2012). Imag- sented at the 10th International Space inative Content and Building Form in Syntax Symposium, London. the Seattle Central Public Library. Pa- Sailer, K., Pachilova, R., Kostopou- per presented at the 8th International lou, E., Pradinuk, R., MacKinnon, D., Space Syntax Symposium, Santiago de & Hoofwijk, T. (2013). How Strongly Chile.

Te dynamics and diversity of space use in the British Library