‘Patterns are Morphological Lava" Fred K. Sehaefor PATTERNS OF LocKTTON William Bunge Schaefer's "morphological laws" are receiving increased attention by theo= reticel geographers. ‘ve explore the subject. Introduction ‘This paper is intended to be a second supplement to the book, Theoretical Geouraphy.> In the baste bok, geography was related to ecience after Christaller's primary substantive example and Schaefer's primary methodology. The central problem of theoretical geography was tentatively identified as "the nearness problem," that is, "to located interacting objects as near to ench other a6 poo— sible." The first supplenent to the basic book is an article entitled "Spatial Relations: The Subject of Theoretical Geography." ‘There the focus vas cherpened by a hard examination of the methodological coments of select contemporary theoretical geographers and by a cystematic review of come of the oathematice of space as applied to geography.2 IL, Thrae Types of Location Predictive or theoretical geography has important old roots in geography; certainly Davis and Réppom vere predictive geographers and n close histories) examination might revoal much earlier antecedents. As in prospecting for oil, so in history, we usually find traces of what we seek if we Look hard enough. Hut the min historical upthrust of the American-Snedish echool of predictive geograply has been through Christaller. Schaefer crystallized the methodological implications of Christaller's and related substantive work, Sehaofer described the essence of predictive geography as the discovery of predictive patterns. Schaefer's insiatence on spatial patterns (or the earth's geometry, or spatial. structure, or whatever term you prefer) seen to run head on into predictive geographers' interest in movements (or circulations or spatial process or what- ever tern you prefer). he history of this iuportant methodological contribution reported elsewhere.? ‘Tentative conclusions are that most locations on the earth's curface are explained, which is logically equivalent to predicted, on the basis of optinal movenents (or geodesics, or least effort paths or whatever tera you favor). Earlier the author ha identified this recurring thene as the nearness problem - the problen of locating interacting objects as near to each other as possible. ‘The catholie elaim ende for this approach 1e that it cuts across all branches of traditional aystenatie geography. Also, it is admitted outright that the probabilistic aspects of predictive geography do not conform to optimal arrangenente." With thie important adwiesion behind us the remainder of this paper witl contain no further comment on lecation theory based on prob- abilities. The Schsefer versus almost all others!' argunent in fevor of patterns rather than movements as the subject of theoretical geography dissolves to nothing vhen it ds realized that particular optinal movements reault in particular patterns, that the geometry and movenents are intertwined in epatial harmoay. In this pase?) We WiLL approach our subject from the point of view of Schaefer's patterns rather than interactions or novewents of ULlnan und others.” In gore ways Schaefer's pattern approach to this @alectic is the core appealing of the two since the patterns are co easy to "map." Of course, the "mapa" are not true maps since they de not apply to any particular locations on the earth's geogran! eurfaco} rather, they are idealized "maps! cuch aa KUppon's Hypothetical Con ‘tinent. Very recent substantive work, especially by Boyce and Clark, Dacey, Nordbeck, and Tobler, is sharply in the spirit of a Universal Systenatic Geography. © Dacoy especially has been engaged in work on patterns. Put-theme basic inspiration for this payer was received from Chriatallor and one of ‘Yobler's favorite nongeographie teachers, Thompson.? Christaller makes two isotropic assumptions; (1) that the density of his base object is uniform (even distribution of rural population) and (2) that his epsce ds not twieted (uniform transportation in all directions). Thompson's work, desling with the spatial but rongeographie aspects of Life forms, contains most suggestive material. The very title of Thompeante beok, On Growth and Form, can be paraphrased Qn Movement and Geometry without too meh violence to hie intended meanings Startling resemblances between geographic problema and those in biology might tempt renders to conclude, using our oun argument of conservation of academic effort, that a really efficient ecience of location would include the spatial aspects of biology, physics, and co forth, as a study in Goners] Systems. I think not. The very range of opatdal scale between various ceience introduces worlds so weird that spatial exports in one have little to say to spatial experts dn another, Subatomic physics ie at a scale eo different from the earth's curface ac the home of man that the space is qualitatively almost totally foreign. Sub- atomic geometries have been superficially apposed to astronomical geometry by comparisons between the orbite of planets and electrons. In spite of carly and prolouged attempts at such synthesis, efforts have been unsuccessful and the persistent experience of failure has been codified as "reductionian.! Howevery while grandiose comparizone have beon general failures some cross inspiration can be expected. For instance, Eigen values obtained from the characteristics of matrices have been used to explein ring Jumps of electrons. In geography we might explore the characteristic of linear transforrutions ag revealed in their matrices to seo 4f Christallei 4n the geometry of ayemstry might shed Light on Christaller's crucial "fixed k!' 2 settlements occur in rings. his exploration assumption. Electrical engineers, biologists, economiste and others vill learn puch fron us, and we from them, but this merely expresses the ultimate universality and interconnection of all knowledge. ‘The logical conclusion to be drawn from successful borrowing ef spatial notions in biology and elsewhere ie not necess sarily the horrer of feoling compelled to claim all spatial problems are geography simply because all geograpiy seems to be spatial. Throughout this paper references are made to nongeographic subjects whore geographers cnn expect both te learn and to teach, bat no Wild territorial claims are made. 2pie Returning to the tain polat of Thompeot's insights, he noticed that one eould accurately meaoure the length of a given specie of fish by menmuring ite weight. He assumed a constant density and chape to the fich. The outline of an object can be considered to be a property of density. Where the object ends ite density suddenly drops to zero, a discontinuity to ite density. ‘Therefore, srnal density and external density (shape) are in fact essentially the single element of density. To @ modern geographer the resenbalancos and differences ens that to Chriatatler are immediately apparent. Christeller avoided the pro the irregular density of chape introduce, that is, he avoided the boundary problem, dy the often used device of imagining an infinite plain, Poth Thompson and Christaller assumed uniform internal density, Christaller aleo ascuned a uniform transportation arface while Thompson simply ignored the internal circulation of his fish. Furthermore, one of the ways in which wa have powerfully extended Christater's “dimensional tension" involved in the problem of trying to spreed pointe over an area, ? So that while Chrictaller higeelf aid not explore the dincnsionsl ginal work is in the direction of understanding the basic problems implicst in his werk, hia intellectual descendents were compelled too. Thus, when Thompson etphasized the dimensional aspects of his fish, another correspondence to Christaller's work 1s established. To translate biology into geograghy, consider the problem of predicting the volume of water at a river's mouth (weight of fish). If the rainfall is sonstant over the valley basine (fish are of uniform density) and the shapes of the basing (fieh) the came, then the lengths of the valleys (fick) are propor- tional to the capacities (weights) of the rivers (fish). In addition, the pattern of the river oystem (veins and arteries) depends on the slope (internal transportability) of the terrain (fishes body). We now identify the cpatial eleuents involved as dimension, morphology and density. The three elements all we highly independent of one another?¢ appear to be mathematical groups and ther At this speculative point in our underetanding it also appears that there is an order to their fundamentality. Disensions see to be the most basic followed by morphology and lastly density. Christaller, Thinen and others exhibit a tondency ta want to reduce the problems to clenental dimensional fora. Geog- raphera ore fairly skilled in shifting dimensions. A dot map of elevation or a dot map of nena annual reinfall hao every bit as much elaim to legitimacy as hidetound isoline representations, but these are transformations within the sot of donsity are mich nore violent and nore rare ‘The truly elemental nature of our discussion strikes howe when we notice ensions. ‘The transformations between dimensions and morphology and/or that the three elements ara actually three fundamentally different ways of5 predicting location, Therefore, I believe, we are at the heart of theoretical geography. A, Patterns of Dinenaional Lecation ia this section of our discussion only dimensions will be allowed to vary. Nerpholegy and density are asauned to be unifora. To obtain more accuracy and cover certain areas in classical climatology and oceanography and new problems in political geography, we could deal with three @imensions (volumes) a8 well as 2ero (pointe), one (lines) and two (areas) but it would merely clutter the argument so we will not inelude three or Wigher spaces Js "Locate points on Lines zo that they are as near to the line as possible! The dual and equivalent statewent is, "irrange points on lines so that they are ac far from each other as pocsible,'! ‘The pattern is = uniform distribution of points along a line. (Figure ls) gure 1, Points space uniformly along a line, The scale of the spacing in this and cubsequent examples depends on the ratios of the objects to cach other and is triviel in term of our dLecussions Notice that if only one point is involved the median center ts the solution; therefore, the Problem can be thought of as one of "miltiple medians. The dual statement suggests an analogue computer based on aagnots, “+ If soall bar magnets are skewered through corks and all the mgnete turned the cane way so as to mutually repel] each other and then tho corked magnets are allowed to float in e long arrow trough of water, they vill form a uniform pattern along a Line. One method of obtaining » grasp of the power of the pattern is to stara at the unlabelled pattern and ask yourself “Of what is this a map?" Some poselble angware inelude filling stations along @ highway, major voleante peaks along the Cascades and the distribution of ice cream vendors along a beach. Natice that these suggestod applications to the earth's surface are more than shallow spatial coincidences, For instance, the total travel eost along a beach for the conaumer ef ice cream is minimized by such « pattern. +® The volosnic pattern mininizes the movenont of mages in the fissure, or put in another way, the uniform dis- tribution marks points of the greatest internal pressure. Since we ere approaching our subject this tise from Schaefer's point of view of patterns, in the remainder of the article we will not yoint out the rather=6- obvious examples of minimized wovementg which each pattern representul sin en area so that the lines are as near to the 2. Locate straight Li area as possible." ‘The dual; "Arrange atraight lines in an aren so that the linee ere as far from each other as possible." Whe pattern is one of straight parallel lines evenly spaced. (Figure 2.) Figure 2. Parallel linea evenly spaced. Straight lines are implied since we are allowing no morphology in our dimensional locations. Again the problem might ve thought of in terms of self= repulsing floating magnets. This time the magnete are all tied together by strings to form celf-repulsing lines, Notice that = ingle line "trying to get away from itseli" isa straight line. The dual atatenent of placing objects ne far from asch other as poceible ie becoming rather tedious and will ue dropped in most subsequent examples, It is well known that any extremum problem can be stated as 8 maximum or a minisum but still the reader might gain insight by conducting the exercise by hinsel?, Viewing the pattern as a map, such obvious translations as ridge and valley topography come to mind, I believe that the parallel Line pattern is the most basic Mne-in-area pattern an] that it da the fundamental river system, railroad pattern, otestera. 3. “Locate points in an area as near to the area as possible." ‘The pattern the equilateral triangular distribution made co famous by Christaller. (Fig. 3a) Figure 3a. Bquilatoral triangular distribution of points.ae Possible maps include a range of objects such as classical sottlenent patterns, distributions of wild animale end efficient arrangements of oil wells. So far ve have draw attention to the dialectic between the geometry and the movements but the dialectic cam be even further expanded. Notice that the covileteral triasgular @istribution formlly implies a hexagonal net. (Figure 3b.) \f vn < pA = \ —<) OF Na f \ \ at Figure 3d. Net of regular hexagons without pointe. That is, we can eotablich the center of each celd in the net and thus generate the equilateral triangular distribution, or wo can ack for the locus of all points halfway between the equilateral triangular distribution and innediately generate the ket of hexagons, Stated in central place terms, tho location of the central incos determines the trade area and vice verea. Tt is really redundent to give both. ‘hese geometric duals appear often and, in combination with the implied tovements, give a rich understanding from little, For instance, in Chrictaller's theory, with the single exogenous variable of k, say equals seven, consider the vich map we can drav. We can place settlements of various sizes, know their range of goede, their market boundaries, the movements to and from the centers te their hinterlands and even a great deal about the structure of their prices. he richness of Christaller's theoretical concepts bave thelr foundations in the various formal mathemmtical dualities. ‘ky Locate (straight) lines as near to points uniformly (equilateral tri- angularly) distributed in an area as possible." The pattern is, not too surprie- ingly, a sot of parallel ovenly opaced lines. (Figure 4.) The points mst be Figure 4. Yarallel lines evenly" spaced among an equilateral’ triangular~8- or they will toke on variable density. since arranged in @ unifora patter one can project a methematical surface into « density surface and conversely, obviously, only a unifora distribution of points will be "flat!" If we coneider just one line notice how closely the problem xeaonbles that of regression. In the first example and in the immediately previous example, the problema cen be thought of as one of finding miltiple medians. Here the problem is one of multiple regression in the senae of cimltancously fitting many lines to the "data." hie last example has introduced objects of three different dimensions at once--lines and points located in en erea, Thore is no reason to stop here. We could have four different kinds of points, say representing farmateads, villages, tonns and cities all simultaneously located as near to each other as possible. What is evidently needed is come notation system for expressing the possible dimensional combinations, Often good notation systems are guggactive of mathoanticn relationehips that further simplify and provide ceoper insight, tut a search for notation would place us ahead of ourselves: Instead, let us continue an informal exploration of the half seen theoretical landatape avirling out of tho fogs B. Patterns of Morphological Lacation Points con not be shaped but toth Lines and areas cans Gone mot still hange ever what We wean by shaped apace, fn area of disuniform transport cost can be thought of as stretched and puckered co that cirelee of oquel cost would not draw on the ares as circles of metric distance, These cireles can be thought of as not just Tiseop's circles which preserve circles in the small in conformal naps, but circles in the large as well co that both enell and large angles, and thug all shapes, are preserved, A true "conformality" in the broadest application of the word, We refer to these apace puckeringe ond stretchings as "internally shaped." Notice, again that lince as well ae areas have the possibility of being internally shaped. Ag an aside, we should be diseaticfied with our over reference to cost-ailos and time-miles and even with Tobler's cophisticated utile-niles,’? Perhapa our almost exclusive concern with such space-warpers is due to the dicproyortionate influence of econonie geegraphy in current theoretical work. We need a griely “death-uiles" distances to explain human migration of @ grees planetary sort. In cliatology there oxicts the space twis ter of Coreolis acceleration, Coreclia~ miles are just as legitimate a5 cost-miles. AlL the previously mentioned pattertis are seriously affected if ve introduce the twieting of space. Obviously, if two paints are nt half the real niles ag compared to the earth milas, they are located twiee az close As an oven phote~= graph would indicate. The reader can readily imagine aany examples but some are oxplicated here because they suom to shed genuine insight of a rather startling sorte 1, "Locate finite areas of different real-miles as near to a single point as possible." the pettern is one of concentric circles whose radii increase in urequal incresents, (Figure 5) ‘Treathd as a aap, Thllnen rings of agriculture Figure 5. Concentric circles whose radii inereave in unequal increnents to produce irregular rings, immediately cone to aiid. ‘The rings suggest ecological cixelos of animale, from frogs to camels around a desert vator hole. ings ef volcanic debris are-caused by varying transportability of expelled materials 2, “Locate finite areas of different real. wiles as near to avstraight Mine as possible." (Figure 6.) ‘hllnea strips, with the cetitral point replaced by a Figare 6. Stripe of wmequal width parallel to a base line, soutrall Line, the most obvious extension imaginable but te my knowledge completely noglacted, in apparent. Besides the strips of agricultural land use along-10- \Aldorness roads, perhaps littoral-using ocean animals are similarly arranged with a band of chort flight birds toward the inside and salmon and seals in the farthest band. Vie have accepted that mich of tha movenent on the earth's gurface can be explained ty least-offort paths. Therefore, the shape of the space, the corpho- Logical pattern to the space, can yield infinite patterns of objects. Tf the space ic symmetrically shaped, the objects often form beautiful patterns. Let us examine sone of the patterns of flow dae to fairly eymietrie apace twisting. he brilliant Beckuaan bas produced several "maps" chowing the flow of economic goods with verioue tuietings, 2* (Eigure 7.) Prager, in the Becknann tradition f | Figure 7, Becknann patterns of flow. Stippled region indicates area of consumption, has produced "naps" of telephone or any-other-type-af-network patterns» 1° (Fige 8.) Se) i Reriion © 7 Moin trunk Feeder trunie:-u- There areeotwral-oxamples that are available to géograrhy from our traditional area of interost in physicel geography. - Figure 9 sows the sattera of wind flow Figure 9. Wind pattern in a high, while discounting earth's rotations ,, je and 10% phow a BM ee ty 10a Contours of Stream Valley Pattern 10b Velocity Vectors - Length of arrows Proportional to clope. typical stream pattern Literally flowing down the potential surfaces Maybe it oc~ curred to traditions] physical geographere that thelr elopes were aerely apace— twisters varying the traneportability of unter. Naybo 4t eccurred to them that the flow of water over the terrain is a vector flew over the potentisl map. But after Lohman in 1874 presented the geographic world with the hechure map, why is it not generally known among goograyhors alcoat | a bundred years later that bie hochures were merely a crude vector repreceutation ef the potential (contour) dual? Flow diagruna such ac streams, are density vactere ae opposad to velocity vectore. Actually the density vectors can be computed. fron the velocity vectors but not the other way around. ‘hat is, Figure 10a is reproducible from Figura 10h but pot froaEigure 1.