None of the organizational changes made in response to the expansion of the park system in the 1930s would have greater long-term ramifications for administration of the Park Service than the establishment of regional offices in 1937. The creation of a new level of administration between the Washington office and the field was not, it must be made clear, a new idea. Park Service officials long had been concerned over the difficulty of effectively supervising and coordinating a widely-scattered system of parks and monuments from Washington, D.C.  During the 1920s the Service had established field offices in Yellowstone National Park, Los Angeles, San Francisco, Denver, Portland, and Berkeley.  These offices performed specific functions--landscape architecture, sanitation, engineering, and education (interpretation) for example--and did not exercise any general administrative or supervisory control over parks and monuments. A more immediate example of regionalization was the system developed to administer the Civilian Conservation Corps described on pages 77-96. In fact, because such a large number of NPS employees were involved directly in ECW, Director Cammerer stated in 1936 that the Service was already 70 percent regionalized. 
At a 1934 park superintendent's conference, held while preliminary discussions regarding regionalization were underway in the Washington office, it became clear that many people in the field as well as in the Washington office believed that the problem of communication was becoming critical as the Park System expanded. Reacting to a suggestion that the Park Service adopt a regional system roughly similar to that already employed by the Forest Service, Frank Pinkley, Superintendent of the Southwest Monuments, indicated that he had become increasingly concerned with the separation between the Washington office and the field, and that of "at least twenty different superintendents" with whom he had discussed the matter, all were of the same opinion.  Speaking for superintendents of the new historical areas, B. Floyd Flickinger of Colonial seconded Pinkley's observations, and indicated that he believed that the greater coordination that would come from regionalization was especially critical for the historical areas.  None of the superintendents spoke out against regionalization at the conference Yet, while those superintendents from cultural areas were enthusiastic over the possibility of regionalization of the Service, many of the superintendents from the natural areas were less so. While agreeing that "anyone in the field for years past must have realized we would have to come to some form of regionalization," John R. White of Sequoia National Park cautioned:
Actually, White continued, a more economical solution to the problem of communication than regionalization might be simply to have the superintendents travel to Washington more often. 
The plan advanced before the superintendents in 1934 would have established as many as five regions determined by classification of areas. Two regions would have incorporated cultural areas--one the Southwestern monuments, the other the military parks and monuments.  The scenic parks and monuments would have been divided among as many as three regions. 
The chief executive for each region would be responsible for overseeing that the policies and principles enunciated by the Washington office were implemented by field personnel. To facilitate communication between the Washington office and regions, one of the regional directors would be required to be in Washington at all times. 
Because of funding problems, it was believed that at least for the short run, the regional director would be a "qualified superintendent." Seemingly, the superintendent who served as regional director would not be relieved of his duties in the park.
Little was done, apparently, to follow up the discussions held in 1934. It was not until January 26, 1936, that Director Cammerer appointed a committee headed by Assistant Director Hillory A. Tolson to study the question of regionalization and submit a plan. 
In mid-February, the committee forwarded to Cammerer a plan of "a simple organization that can be manned and administered from trained personnel and money now available."  The regional system proposed would, the committee said, bring the director and his assistants back into a more intimate touch with the field. It would allow greater supervision of the field, while preserving the autonomy and individuality of the parks. Administrative decisions could be made in the field rather than in Washington, and because the proposal would strengthen the influence of professional branches, those decisions would be based on the best technical advice. The system would, finally, provide greater channels of promotion from park to park, parks to region, and regions to Washington. Because promotion opportunities would occur in the various branches, an individual could advance within his profession, and not be necessarily diverted into administration .
The memorandum discussed above did not spell out the make-up of regions. That came several days later. The proposed system was based on a combination of unit classification and geography similar to that suggested to the superintendents in 1934. 
Region 1, with Chief Historian Verne E. Chatelain as the recommended regional director, would include all historical and military parks, monuments, battlefield sites, and miscellaneous memorials east of the Mississippi River.  Region 2, the second region established primarily on a classification of areas would have been headed by Frank Pinkley, Superintendent of the Southwestern monuments. Pinkley's region would have included the southwestern monuments as well as Mesa Verde and Carlsbad Cavern national parks, and Petrified Forest, Wheeler, and Great Sand Dunes national monuments.
The remaining three regions would have been headed by superintendents of large natural parks--C.G. Thompson of Yosemite (No. 3), Superintendent O.A. Tomlinson of Mount Rainier (No. 4), and Roger Toll of Yellowstone (No. 5).  The primary division was geographical and the regions would have included both natural and cultural areas:
It was believed that eight existing and projected parks--Acadia, Great Smoky Mountains, Platt, Hot Springs, Isle Royale (projected), Mammoth Cave (projected), and Everglades (projected) could function as they did for the present, although it was recommended that when Mammoth Cave and Everglades were established they would be coupled with Great Smoky Mountains, Platt, and Hot Springs into Region 6.
On March 14, Acting Director Arthur Demaray, forwarded a memorandum that described in detail the responsibilities of the proposed regional offices. Duties and responsibilities would certainly change over time, he said, but the following were representative of the work that it was proposed to transfer:
The Service would go slowly with regionalization, Demaray concluded, and would enlarge the authority of the regional directors only when such action was justified by experience. 
Secretary Ickes answered Cammerer's request on March 25. Reflecting his well-known antipathy for bureaucracies, Ickes wrote that he was reluctant to agree to the creation of offices outside Washington "because it would be only a question of time until a bureaucratic field force would become established to the detriment of the Washington office." While he recognized that Washington officials had to be "fully informed of administrative problems and actions," he did not believe that creation of regional offices would contribute to that end.  Rather, he believed that the appointment of district supervisors in the Washington office would be more effective, and instructed Cammerer to revise the proposal accordingly.
Ickes was not the only one to express reservations regarding regionalization. A flurry of letters to the secretary, which Cammerer believed was inspired by the National Parks Association, all indicated a concern that the grouping of historical and natural areas would be to the detriment of the latter.  Within the Service many "old-line superintendents object to the concept as an unwarranted intrusion on their ability to communicate directly with the Washington Office, and many rank and file personnel saw it as a barrier to career advancement." 
Director Cammerer believed that much of the opposition to regionalization from both groups would be dissipated by appointing "old-time" Park Service men to head the various regions.  While opposition to regionalization did not immediately disappear, Cammerer and his deputies were able to blunt the efforts of it, and convince Secretary Ickes.
On January 21, 1937, more than two years since regionalization was first discussed at the Annual Superintendent's Conference, Secretary Ickes initialed his approval of a regional system that would be implemented after the end of the fiscal year. 
Accordingly, on August 7, 1937, Director Cammerer issued a memorandum that implemented regionalization of the National Park Service. The plan approved by Secretary Ickes established four geographic regions:
Interestingly, of the first four regional directors only one, Thomas Allen, Jr. (Region II), had been a superintendent of a natural park, although Carl P. Russell (Region I) and Frank Kitteridge (Region IV) had considerable National Park Service experience.  Herbert Maier, who was named acting director of Region III, had been in charge of the Service's CCC and emergency activities of Region III (CCC). In addition, the associate regional director would be the current CCC regional officer. 
The implementing memorandum made it clear that the Washington office intended to proceed cautiously with regionalization, and subsequent memorandums issued throughout the rest of the decade amplified, refined, or in some cases altered functions of the regional offices.  Nevertheless, the outlines of the organization that would administer the National Park Service in the future were drawn, and it reflected Secretary Ickes concern that the field offices not rival the Washington office:
Establishment of regional administrative units was an experiment. Within a short time it proved its effectiveness to both Washington officials and field personnel. In his annual report of 1938, Director Cammerer wrote:
The following year, while calling for establishment of one additional region, the park superintendents resolved:
Chapter Six continues with...