This booklet illustrates the use of Geographic Information Systems to examine the impact of industrial technology on natural and human environments, using as an example railroads in

England during the nineteenth and early twentieth centuries. It presents GIS methods as a fruitful combination of computer-assisted mapping and data analysis. The text introduces techniques of statistical and spatial analysis to search for patterns and to answer the two basic questions of a GIS approach: What is where? and Why is it there? The problem examined is one in social and environmental history. In England and Wales during an era of intense industrialization and urbanization, what patterns of population change can be identified and how can those changes be explained? What was the timing, magnitude, and geography of urbanization and rural depopulation? And how was the extension of the railway system related to these changes? To what extent did the extension of rail system into rural areas accelerate--or retard--rural out migration?

Although such "classic" questions have been studied often before, the use GIS methods opens new ways of taking them up. Working with desktop GIS can greatly enhance students' interest in and understanding of quantitative aspects of problem solving. With a flexibility and speed unavailable to historical geographers and historians before now, they can map and study changing quantities in geographic space and gauge the statistical and spatial relationships among a number of variables. In this study, the variables of chief interest concern the attributes of census Registration Districts in England and Wales from 1851 to 1914: population density, the extent of rail-line coverage, and the rate of migration among groups defined by age and sex.

Ideas and Concepts

The social and economic transformation of nineteenth-century England and Wales is the classic example of western industrialization and urbanization. Viewed from the perspective of social and environmental history, this transformation provides an interesting way to examine the impact of new technology on past human and physical environments. One far-reaching example is the steam-powered railroad system which, from its beginning in the 1830s to its apogee on the eve of World War I, grew to reach nearly all corners of England and Wales. The landscape of the Victorian City was a monument to the railway age, with its huge train stations and rail yards, together with the great earthworks and tunnels that the rail network required. And to its stations came more and more individuals and families who were moving to town in search of better opportunities, leaving the countryside behind and villages in decline.

At first glance, the argument that railroads facilitated internal migration, the decline of the rural communities, and the growth of cities seems a truisim, something hardly disputable. A closer look at the literature and history, however, suggests that this argument bears re-consideration. Although building of railway plants, junctions, and freight yards increased the populations where these works were located, the notion that railways were a primary cause of urban growth and rural depopulation has proved dubious and overly simple. As recent research has shown, new rail lines into major urban centers usually arrived after the population had grown significantly and not before. With one or two exceptions, railways in England did create new towns as in the United States.

Did the railways facilitate migration from countryside to town? A look into the relevant literature on the subject shows little systematic work on the question. All studies agree that the Victorian and Edwardian eras saw the rail system reach its maximum extent and popularity as means of transporting people and freight. But if this was the Golden Age of the railroads, how they affected the movements of people from village to town and from town to city is more often a matter of conjecture than demonstration. Even less attention has been given to another question that we can address. To what extent did the extension of the rail system into small towns and villages retard the movement of people from small settlements to large urban centers or to regions of greater economic opportunity, such as the coal-mining districts of southern Wales? Rather than hastening a rural exodus, it is possible that the coming of railways and the commercial opportunities they brought opened new jobs for local people and thus served to stem the pace of rural out migration.

The use of GIS makes it easy to take up other questions as well. For example, did men and women migrate in similar patterns? Or did men and women tend to move to different destinations because opportunities for work varied by region and by gender? What was the timing, the extent, and geography of rural depopulation? When and where did the peak of the rural exodus occur?

 

Analysis: Searching and Describing Patterns

What follows here are two examples for the main part of the manuscript. The first is an introductory analysis; the second, an advanced one.

Example 1: Patterns of Migration from 1851 to 1861

This section introduces 1) count and ratio variables (population, population density, net migration, and percentage population change due to migration); 2) the statistical concepts of center, and spread; and 3) their use in defining groups for choropleth mapping. To build understanding, I present a 5 percent random sample of registration districts containing migration data for England and Wales. Using a small number of cases students perform hand calculations and acquire experience with the variables and their interpretation. In the text I use the data for 15 census registration districts--about half of the 5 percent sample--to illustrate the concepts, alternative methods of classifying attribute data into groups (equal intervals, mean and standard deviation, quantiles, and "natural breaks"), and the reasons for choosing quantiles or natural breaks over the two other choices. Two exercises ask students to generate results using the other half of the sample data. (See the attached lab exercises.)

Two sets of maps illustrate the identification of patterns through visualization, that is, the plotting and examination of migration rates for men and women over the decade 1851 to 1861. The first map displays the results from the sample, showing the percentage change over the decade by gender that is attributable to migration. The distribution of the sample indicates areas of out migration in rural areas and in migration in urbanizing areas, but differences between men and women seem negligible, or "too close to call."

A second set of maps displays migration data by gender for the entire set of registration districts for the same decade, 1851-1861. These maps confirm the patterns of out migration and in migration that were evident in the sample, while revealing the greater complexities. The differences between male and female patterns are more pronounced and suggestive. Whereas male migration proceeded to a limited number of destinations where industrial and mining work was concentrated, female migration was geographically more extensive. During this decade at least, women migrated in significant numbers to a broader range of destinations, some moving to districts favored by men and others to the hinterlands of industrial zones and of London especially. These interesting patterns also suggest--but do not demonstrate--that women were likely to migrate over shorter distances than men, reflecting the more dispersed availability of work in domestic service, the largest segment of employment for women. A methodological lesson of these comparison is both the utility and limitations of a sample as compared to the population of which it is a part.

[Map 1 about here]

 

 

Example 2: Railways and Rural Out Migration

This example tests my hypothesis that the extension of the rail system into rural England Wales from 1850 to 1914 retarded--rather than accelerated--rural out migration. The results provide some confirmation as well as further questions to explore.

After presenting maps showing the extension of the rail system from 1845 to 1914, I divide rural registration districts into two groups, one with low access to rail transport (those that fall into the lower quartile of rail line density) and the other group with high access (those falling in the upper three quartiles of rail line density).

[Maps 2 and 3 about here]

 

A t-test of changes in population density in these two groups of Registration Districts during the 1870s shows a small but statistically significant difference between the two groups. From 1851 to 1871 the mean population density declined in rural districts with low access to rail transport, while it remained unchanged in districts with greater access to rail. Results for the period 1881 to 1914 confirm the pattern. Although average population density for both groups of districts declined, the decline was significantly larger in the districts where access to railways was lower.

One caveat: at this point I'm not certain whether to include the results of a t-test because of the additional space required to introduce the test to an undergraduate audience with presumably little or no background in statistics. I would prefer to demonstrate the relationship via a map and visualize examination, if I can improve upon my first efforts at displaying two variables on a map. (See Map 3.) I would welcome suggestions or reactions here.

Table 1. t-Tests of Changes in Population Densities, 1851-1871, and 1881-1891

Group 1: Districts in lower quartile of rail density (1914)

Group 2: Districts in upper three quartiles of rail density (1914)

The purpose of this exercise is threefold: to gain understanding of the statistical concepts of distribution as well as the center and spread of a distribution; to familiarize yourself with two variables that measure migration; and to understand the difference between scales of measurement using counts as compared to rates or ratios. In this exercise, use the data in the spreadsheet on migration from 1851 to 1861 for fifteen, randomly sampled census registration districts of England and Wales.

1. Create a dot plot. On a piece of graph paper, create a display that shows the distribution of values for the variable "Total Migration" for the decade 1851-1861. Start by drawing a line that encompasses the minimum and maximum values. Then place dots on the line to indicate the positions of each value in the distribution. Your line should look something like this, with a zero near the center to permit your plotting of positive and negative values and a proper label.

 

---

Min

0

Max

 

Dot Plot of Total Migration 1851-1861

Recall the definition of this variable: for each census Registration District, the change in population from 1851 to 1861 that is attributable to migration. Negative values indicate a decline in the size of the population due to (out) migration; positive values, an increase due to (in) migration.

2. Briefly describe the distribution of values for each variable. What is the range of the distribution, i.e., the difference between the maximum and the minimum? Are the values spread out evenly? Do they cluster?
3. In a sentence or two describe what you think your plot indicates about the pattern of migration during the decade 1851 to 1861.
4. Create a table showing the Center and Range of the distribution you just studied. Calculate the mean, the median, and the range for the same variable, i.e., "Total Migration" for the decade 1851--1861 among your sample of districts

 

Decade	Mean	Median	Range
1851-1861

 

 

5. What do these measures indicate about the distribution and change of the population in your sample districts that was due to migration?
6. Are the mean and the median similar or different? If different, what is it about the distribution that accounts for the difference?
7. If the mean and median are different, which statistic would you use to indicate the "average" or "central" value of this distribution? Why? (Hint: consider the effect of extreme values.)
8. Using the statistic you selected in b), write a sentence or two that describes what the "average" or "central" value and the range indicate about the pattern of migration for the decade 1851-1861.

Standardization: From Counts to Ratios/Rates

9. Counts. Return to your dot plot showing the distribution of total migration from 1851 to 1861. This variable is records the number of individuals who were added or subtracted from the population in 1851 by means of migration into or out of a district. It measures this change by recording counts of individuals.

Place this plot of counts next to the spreadsheet that contains all the sample data. Choose one of the larger values on the dot plot; then locate on the spread sheet the registration district which has this value. Now, compare the value of Total Migration with the Total Population in 1851. Repeat this comparison for at least five values of differing size.

10. What, if any, pattern seems to emerge from this comparison? Do the large values of Total Migration tend to correspond with large values of Total Population? Or do the large values of Total Migration tend to correspond with small values of Total Population?
11. On the basis of this inspection describe in a sentence or two the limitations of using the count variable Total Migration to compare and gauge migratory movements across your sample of Registration Districts.
12. Rates or Ratios. For each Registration District in your sample, calculate the percentage change in population due to migration from 1851 to 1861. [(Total Migration/Total Population in 1851) x 100]. This new variable records a rate or percentage based on the change due to migration in relation to the size of the population at the beginning of the decade of observation. As you discovered above there is considerable variation among the sample Registration Districts in terms of their population size in 1851. This variation makes it difficult to make a judgement about the significance of any given population change due to migration because it's more than likely that districts with small beginning populations will go hand in hand with small changes due to migration, while the opposite is likely for districts with large beginning populations. The use of percentages , This has the effect of creating values which are measured by the same "standard," the size of the population in 1851. tStudy the values and explain in a sentence or two why the conversion of counts to rates/ratios facilitates your ability to compare and gauge migratory movements across the Registration Districts in your sample.
13. Create a dot plot showing the distribution of the new variable that you've created above: the percentage change in population due to migration from 1851 to 1861.
14. Compare this dot plot with the dot plot of the count variable Total Migration which you constructed in step 1. Describe the pattern you see in the distribution of the percentage change due to migration.
15. Explain why the variable measuring the percentage change due to migration facilitates the task of comparing and measuring variations in migratory movements across census Registrations Districts.

Measuring Migration from 1851 to 1861: Grouping Values and Choropleth Mapping

Lab Exercise

Grouping values in a distribution for Choropleth Mapping

1. Using the dot plots and statistics you created in the previous exercise, examine the plot for the percentage change in population due to migration from 1851 to 1861.

---

Min

0

Max

2. On the same sheet of paper, construct 4 (four) copies of this dot plot. (Hint: use a photocopier.)
3. Defining Groups by Equal Intervals. Create five groups of values by dividing the range into 5 equal parts. Draw vertical lines on the first dot plot to designate the five groups defined in this way.
4. Examine each of the groups displayed on the plot. What do you find striking? How many points are in each group? How well does this scheme indicate varying levels of migratory movement?
5. On the blank map of sample Registrtion Districts, construct a choropleth map showing the results of the grouping scheme. First devise a pattern or shading scheme to designate each of the five groups. Apply this scheme to the map.
6. What does this map reveal about migratory movements? What does it conceal?
7. Defining Groups by measures of Center and Spread. For the Registration Districts in your sample, calculate the mean, the median, and the range for the same variable.

 

Percentage Change in Population due to Migration 1851-1861

Decade	Mean	Median	Range
1851-1861

 

8. On the second dot plot, add points to indicate the position of the mean and the median.
9. Groups based on the Mean. Using the mean as the central value, create 5 groups. Start by creating a group centered on the mean; then create two additional groups above the mean and two additional groups below the mean. Use your intuition to create these four groups but be able to explain the reasoning that led to your decisions. (Optional: calculate the standard deviation of the mean, and use the SD to create the four groups surrounding the "mean" group.) Draw vertical lines to designate the five groups on the second dot plot.
10. Examine each of the groups displayed on the plot. What do you find striking? How many points are in each group? How well does this scheme indicate varying levels of migratory movement?
11. On the blank map of sample Registrtion Districts, construct a choropleth map showing the results of the grouping scheme. First devise a pattern or shading scheme to designate each of the five groups. Apply this scheme to the map.
12. What does this map reveal about migratory movements? What does it conceal?
13. Groups based on the Median. Using the median as the central value, create 5 groups. Start by creating one group centered on the median; then create two additional groups above the mean and two additional groups below the mean. Use your intuition to create these four groups but be able to explain the reasoning that led to your decisions. (Optional: calculate the values of the 20^th percentile, the 40^th percentile, the 60^th percentile, and the 80^th percentile. Use these values to create the five groups centered about the median, which is the 50^th percentile). Draw vertical lines to designate the five groups on the third dot plot.
14. Examine each of the groups displayed on the plot. What do you find striking? How many points are in each group? How well does this scheme indicate varying levels of migratory movement?
15. On the blank map of sample Registrtion Districts, construct a choropleth map showing the results of the median or quantile grouping scheme. First devise a pattern or shading scheme to designate each of the five groups. Apply this scheme to the map.
16. What does this map reveal about migratory movements? What does it conceal?
17. Groups based on "Natural Breaks." Examine the distribution of values on the fourth dot plot. Look for clusters of values that seem to suggest "natural breaks" in the distribution, or "natural groups" that emerge from inspecting the distribution. From your visual inspection, create five groups, and draw vertical lines on the fourth dot plot to designate them.
18. Examine each of the groups displayed on the plot. What do you find striking? How many points are in each group? How well does this scheme indicate varying levels of migratory movement?
19. On the blank map of sample Registrtion Districts, construct a choropleth map showing the results of the median or quantile grouping scheme. First devise a pattern or shading scheme to designate each of the five groups. Apply this scheme to the map.
20. What does this map reveal about migratory movements? What does it conceal?

Conclusions

21. In a sentence or two summarize important conclusions that you can draw from this exercise by responding to the following questions.
22. Given the distribution of values for the percentage change of population due to migration from 1851 to 1861, which method of grouping seems preferable to create a choropleth map that highlights the variation of migratory movements across census Registration Districts? In sum, which of the four maps is best for this purpose?
23. If one had a different purpose in mind, it is possible that one of the other maps would be preferable for displaying the point you wish to make. Explain why one of the other maps might be preferable for a different purpose or point?
24. What rule of thumb can you suggest to guide the construction of a meaningful choropleth map of an attribute/variable.