geo referencing a historic topo map tif to the correct geographic coord system

[KenLucas]

I have a 1910 topo map tif that I want to geo reference, using the lat & long coordinates listed in the 4 corners of the map tif. I realize I can just set the arcmap projection to WGS84 to get lat & long, prior to geo referencing. However, I'm wondering if I should set the arcmap session to an older geographic coordinate system. Alternatively, is there a projected coordinate system, such one in conjunction with NAD1927, that would be more appropriate?
   I geo reffed the tif, using the WGS84, compared the PLSS section corners of the historical map tif to the current section corner PLSS position, and determined a 1,200 discrepancy in the section corner's position.
I'd appreciate any comments on this topic,
Ken

[JimCousins]

Hello, Ken,
WGS84 is certainly a mistake. The datum prior to NAD27 was NAD (north american datum), and it was based on Clarke 1866 spheroid. I feel that georeferencing by features to a newer (with a specified datum) map is a good solution, providing there are major features to do this with.
The issue with this is these maps are not that accurate to start with. They were typically created with field survey and plane table geometry, and heavy use of cartographic license.
Perhaps Melita Kennedy of ESRI can correct me / shed some more light on this.
Best Regards,
Jim

[RichardDaniels]

Just as a point of clarification, most field surveys conducted with "plain table" were highly accurate. If the map you are using was created by the USGS or Coastal Survey (or their predecessors) the data will be of high quality. Your issue is selecting and using the best method for georeferencing your information. If you are georeferencing a topo map you should check to see if you have any benchmarks listed. You may be able to find their current coordinates and use them to georeference you map as well.

For step by step method for georeferencing historical maps take at look at http://proceedings.esri.com/library/userconf/proc98/PROCEED/TO300/PAP300/P300.HTM

[KenLucas]

Jim,
   You've really been a big help to me in past threads. I'm really grateful for your response.
   To clarify: my team is attempting to get the PLSS lines used in 1910, into a geo referenced tif, so that we may geo reference some other old maps to the old PLSS tif. For these 1910 map tifs, we're not using map features for the 4 link points, rather we are using the lat and long coordinates found in each of the 4 corners of the maps. We are simply just entering the lat and long coords, printed on the map, into the link x y input box. I get a 1,000' discrepancy between the position of the old map tif section corners and new PLSS corners. I expected discrepancy, but does this seem reasonable?
I wanted to use the nad27 datum, but I don't know how to get the arcmap's layer properties into lat and long units, using the NAD27 datum. Based on this additional info, what would you try?
  You gave me a good idea: my team is hosting me to the San Diego conference in two weeks. Priority should be to track Melita down, if she is there, and show her a sample of my data. I'd like to meet you, also. Are you going to make the conference? If so, I owe you lunch.
Ken

[KenLucas]

Jim,
  I apologize. I didn't comprehend your response closely enough. I just changed the arcmap properties coordinate sys to Clarke 1866, as you suggested, and the map units are now in lat and long. Is this what we should use, if we enter the lat and long cords into the link x y input? Or would it be more effective to link to map features?  We realize there is error in these maps, we're just obligated to simulate the past PLSS as best we can.
Ken

[MelitaKennedy · Accepted solution]

What Jim said.

According to Jan van Sickle, your topo map predates "NAD" by a few years, but the previous datum also used Clarke 1866, so using NAD27, would be a closer fit than WGS84. The map itself may be using the polyconic projection. I don't think I've ever gotten good answers on what parameters to use with polyconic maps, though. If the latitude lines look symmetric, try using the center of the map and project the lat/lon corners to it, then use those to georeference the map.

Melita

[KenLucas]

Melita,
  I appreciate you and Jim for advising us. I'm still unclear of the procedure you are suggesting for setting up the arcmap layer properties, before I bring in any data. My choices seem to be:
1. select Clarke 1866, out of the geographic coord sys folder
2. select NAD27, out of the  north America - geographic coord sys folder
If one of these is not right would you, please, specify how I set up the layer props coordinate system?
I like your suggestion to project lat and long coords to the, less distorted, center area of the old map. I assume I should determine the lat and long for 4 link points near the center of the map before geo referencing.
Ken

[KenLucas]

I changed the coord sys of a blank arcmap session to geographic - north america - NAD27. I added the historic tif in - I projected the lat and long coordinates to 10 points around the perimeter of the tif & 2 in the center - I used those points as geo ref link pts, entering the lat & long cords into the input boxes of the link2. I added in a more recent topo and compared positions in the geo reffed historic toop with the same positions in the new topo. Now I measure about 300-400' discrepancy across the map, instead of the 1,000' error I got when I set the arcmap up with WGS84 coordinates, before geo reffing the historic tif. How can I tell if the discrepancy is due to error within the old PLSS and not just error between the geo ref tif and the new topo data? What would you expect?  My goal is to be able to use this geo reffed historic tif to geo ref some historic property parcels. 
Ken

[EarlKlug]

Thanks having some of the same issues with the various old maps sites that everyone's suggesting or linking too.  I'm finding that my RMS error is .03 yet SD / MT border is off as is South Dakota / Missouri when trying to GeoRef 1872 USA Atlas / 1874 MN atlas for a historical map display.  I enjoy the challenge and maps available on David Rumsey or others maps just a projection or default suggestion would be appreciated.