Tuesday, March 12, 2013

US Census of 1840 - Manufacturing

The data used to create the county mill density map (Figure 1) in Natural Streams and the Legacy of Water-Powered Mills (Walter and Merritts, 2008, DOI:10.1126/science.1151716) was compiled by Franklin and Marshall College students Sauleh Siddiqui, Caitlyn Lippincott and Adam Ross during the summers of 2003 and 2004. Their source material was a PDF copy of Compendium of the Enumeration of the Inhabitants and Statistics of the United States, downloaded from the census.gov website at: http://www.census.gov/prod/www/abs/decennial/1840.htm (now no longer available). They transcribed data for 21 states and 872 counties, for kinds of manufactures that would have been water-powered at that time, including:
  • Flour mills
  • Grist mills
  • Saw mills
  • Fulling mills
  • Powder mills
  • Oil mills
Our data includes the most commonly occurring types of manufactories, whereas the Sixth Census included additional data on other types of manufactories that would have used water-power and that would have had local or regional importance (i.e. forges and rolling mills, woolen manufactories, cotton manufactories, and paper manufactories).

Since the time of our effort, improved digital copies of the census documents have become available elsewhere ( see http://books.google.com/books?id=4zhRAAAAYAAJ ). Also, significant efforts have been undertaken to make available a wide range of historical census data (see below).

While we are making our data available for those who wish to use it, these newer sources should supersede the data hosted here. To download the data visit the data download page.

Other Sources of Data

Table Data

NHGIS

  • Visit the Data Finder page
  • Filter by year 1840
  • Select table name "Type of Establishment"

ICPSR

Note that you must be with a member institution to obtain the data. Documentation for the data is available to any user.

County Boundary Data

NHGIS

NHGIS county boundaries have a common key with which to join NHGIS table data. As explained on the NHGIS website, these boundaries are from a combination of TIGER/LINE data of the US Census, and boundaries from Map Guide to the U.S. Federal Censuses, 1790-1920, by William Thorndale and William Dollarhide.

Newberry Atlas of Historical County Boundaries

Newberry Atlas county boundaries are more detailed and better reflect temporal changes in the names and shapes of counties. These may be more useful for larger-scale mapping. The Newberry Atlas data is a timeseries containing overlapping boundaries, so you will have to select boundaries for the point in time you wish to map. The Sixth Census began on June 1, 1840, so you would query the dataset using the WHERE clause:
"START_N" <= 18400601 AND "END_N" >= 18400601

HUSCO

We used HUSCO boundaries in our original figure and mill density calculations. These are available for purchase from Louisiana State University Geoscience Publications.

Wednesday, March 6, 2013

Mapping Mill Dams from Historic Atlases

The Historic Sediment and Geomorphology Research Group at Franklin & Marshall College has digitized dam locations from a number of historic atlases. The effort began with the work of students conducting independent research projects in South-Central Pennsylvania. Sauleh Siddiqui, Caitlin Lippincott and Adam Ross digitized the H.F. Bridgens, 1864, Bridgen's Atlas of Lancaster County, Pennsylvania during the summers of 2003 and 2004. Zain Rehman digitized the Beach Nichols, 1876, Atlas of York County, Pennsylvania during summer 2006.

This is an introduction to our current approach to digitizing and mapping historic industries, including mills and mill dams, from historic maps. The resulting data inform us about the great intensity of industry in our region during the mid-19th century, and we use the data to help guide our field research into the impacts of that industry on valley-bottom streams and wetlands.

Interactive maps and data downloads are available at the data download page on this website.

Overview of historical maps

Nineteenth century county maps are typically drawn at scales of 1:63,360 or smaller, and include information about landowners, important industries, roads, railroads and significant streams. The mid-nineteenth century saw the advent of more detailed mapping in the form of county atlases wherein each map sheet would contain a township or district map (typical scale of 1 inch = 200 rods or 1:39,600) and perhaps a borough or town (typical scale 1 inch = 20 rods or 1:3,960). At these relatively large scales the cartographer could incorporate additional detail such as the ponds and races associated with water-powered industry, and topographic information in the form of hachures to indicate slope and slope aspect. The best county atlases include an astonishing amount of detail as shown in the table and images below.
Political DivisionTypical ScaleFeatures DepictedDepiction of Industry
State<1:63,360rivers, main roads, towns, important industriessymbolization of industries
County1:63,360streams, roads, towns, many industriessymbolization of industries, some races on streams depicted
Township1:39,600smaller streams, roads, paths, towns and villages, most industries, residencesdepiction of industries, including races and ponds
Town or Borough1:3,960detailed block mappingdepiction of industries with pond shapes, systems of races, and diversion dams serving industrial buildings

Northern Baltimore County, Maryland. Scale 1:63,360. From George Kaiser (1863), Military Map, Baltimore Co., Md. Available online: http://hdl.loc.gov/loc.gmd/g3843b.cw0254500
The town of North East in Cecil County, Maryland. Scale ca. 1:46,500. From S.J. Martenet (1858), Martenet's Map of Cecil County, Maryland. Available online: http://hdl.loc.gov/loc.gmd/g3843c.la000290

The town of Bellefonte in Centre County, Pennsylvania. Scale 1:3,600. From Beach Nichols (1874) Atlas of Centre County, Pennsylvania.

Working with historical maps

Scanning and georeferencing

Historical maps and atlases do not have the same spatial or positional accuracy as do modern basemap data, and so it is not always worthwhile to spend the time to scan and georeference historical maps. Simply reading a paper copy of the historical map and digitizing features in the depicted location might be the most straightforward method, particularly when the features of interest are points. The person who is doing the digitizing would look for key relationships such as position of features of interest relative to important road intersections, bends in streams and stream confluences. If there are relatively few features to digitize then map scanning and georeferencing might not be necessary.

If there are many features to digitize, or if some of the mapped areas have changed significantly, then scanning and georeferencing the historical map can reduce the number of such decisions that must be made. Points on the georeferenced map may be digitized quickly and then moved into more appropriate positions relative to features in the modern basemap.

The decision about which route to take will depend on the number of features to digitize and the scale and positional accuracy of the historical map.

Feature identification

Mills and mill dams

At small map scales that are typical of state maps and county maps the most important mill locations are indicated by labels or symbols that resemble a mill stone (lines radiating away from a circle). At intermediate map scales county maps or county atlas maps might also depict the races associated with a mill or water powered industries. The longest races could exceed one mile in length, and would have been barriers to travel for those traveling overland. The best county atlas maps depict mill works and industries in great detail, showing locations of dams and ponds with the layout of races and mill buildings that these impoundments served. It is possible to use any of these kinds of maps to locate mill dams with varying levels of confidence. The kinds of cartographic representation are listed below (from least certain to with their meaning with respect to dam location:
  • dam implied by mill—the depiction of a mill implies the presence of a dam, but existence and location of a dam are uncertain
  • dam implied by race—the depiction of a race implies the presence of a dam near the upstream end of the race
  • dam implied by pond—the depiction of a pond, with or without a race, implies the presence of a dam near the downstream end of the pond
  • dam shown—the map depicts both pond and race which clearly establishes the location of the dam

Mines and quarries

Mines and quarries are frequently depicted with hachures pointing inward from the edges of a pit. Mines and quarries, although generally less prominent than mills, were important features in shaping the landscape. In many instances the early forms of these industries were in and around streams, diverted streams or mined material in stream beds (placer deposits). These industries also used waterpower to drive belts that lifted and moved ore to the surface or pumped water from underground mines. The dams and ponds associated with these operations are not always shown on maps, but it is useful to be aware of potential impact from mining and quarrying and the dams that provided waterpower to these industries.

Forges and furnaces

Forges and furnaces are common to iron-ore producing regions. These operations used water power to drive trip hammers and bellows as they heated and refined ore into cast iron and wrought iron.

Canals

Canals typically parallel a waterway but these features also required diversion and storage of water in order to fill the system of canals and locks as needed. Dams are often associated with feeder ponds on tributaries to the waterway along which the canal flows.

Shops and factories

Whether a shop or factory is depicted on a historical map depends primarily on the map scale and the significance of the business. These kinds of businesses manufactured a wide variety of items including cotton duck fabric, woolens, shoes, rakes, farm implements and most any product of the time. Many towns had industrial areas that were served by networks of races and dams that branched off of trunk streams.

Other

Nineteenth century cartographers denoted other common features of the time with abbreviated names and labels. The following is a partial list of these abbreviations.
  • B.S.S.—black smith shop
  • W.S.—wagon shop
  • G.&S.—grist and saw

Feature attributes

Features taken from historic maps and atlases may be described using just a few attributes:
  • Latitude—decimal degrees of latitude relative to NAD83
  • Longitude—decimal degrees of longitude relative to NAD83
  • RelatedFacilityLabel—the name of the mill served by a given dam, from the historic map
  • StreamLabel—the name of the stream on which the dam was located, from the historic map
  • RelatedFacilityType—the type of manufacture at the mill, from the historic map in a semi-colon separated list of facility types such as "grist mill; saw mill"
  • Representation—indicates how the map depicts a dam and its relation to a mill; values are: dam shown, implied by pond, implied by race, implied by mill
  • StreamName—the modern name of the stream, from 1:24,000-scale USGS Hydrography
  • SourceMap—citation information for the historic source map or atlas
  • SourceMapYear—the printing year for the historic source map or atlas
  • County—the county name
  • State—the state name
These attributes capture most of the information from the historic map and enable further research.

Refining feature location

Locational information extracted from historical maps can be refined by comparing the drawn cartographic representation of features to modern GIS data such as modern basemaps, orthophotography, and high-definition topography.

Digitized features should be repositioned in ways that are consistent with features that are apparent in both the old map and the new map. By studying both maps it is usually possible to position features correctly relative to streams and roads. Care should be taken, however, to look for relationships that are least likely to have changed over time: streams may have been diverted and roads may have been realigned. This practice improves data quality and narrows the search area for eventual field identification. Once a feature of interest has been field identified then its location may be more accurately fixed.