Given the local sensitivities, it was politically prudent to include both Haleakala and Mauna Kea as potential sites for the new telescope. Also we had no idea, and nor could anyone tell us, how people could be expected to perform at the high altitude of Mauna Kea. Certainly at that site the project would be a lot more costly since, in contrast to Haleakala, it lacked any electric power supply and such access as existed was extremely primitive. The hostile weather on Mauna Kea - its high winds and heavy snowfalls - further detracted from its appeal and added to the cost both for construction and subsequent operation. Clearly it would have to offer a distinctly superior observational quality to compensate. Given the time constraints, however, we could afford no more than 6 months to make a decision between the two sites.
Many factors, some more or less obvious (like cloud cover), enter into the selection of a site for a major telescope. For us there were two of special importance on which relatively little comparative data was available - the attainable image quality (the ‘seeing’) - and the infrared transparency which is critical in determining how incisively one may study the cooler bodies in the cosmos which include, but are certainly not limited to, solar-system objects.
Two main methods were in use at the time for assessing seeing. The first measured the random motion of the image of a star formed by a small testing telescope. The second looked at the diffraction characteristics of a stellar image which in fine conditions appears (under high enough magnification) as a set of (diffraction) rings centered about a strong bright core. Variations in the overlying atmosphere, which cause the image to move around randomly, blur these rings until they disappear altogether in poor seeing. An index of the seeing quality is obtained by counting how many rings one sees (if one sees any at all) and assessing how steady they are. Using the telescope which Kuiper had erected on Puu Poliahu, his associates Alika Herring and Bill Hartmann had observed such diffraction rings over a period of several months in 1963/4.
Scientists at the Kitt Peak National Observatory had used a telescope pointed at the pole star to measure the size of the image spread in testing the quality of their sites. The problem I had with that procedure (apart from the fact that Polaris was low in the sky in Hawaii, which would confuse the interpretation) was how to differentiate the image motion caused by atmospheric inhomogeneities (which we wanted to measure) from that due to shaking of the test telescope which, of course, had nothing to do with the inherent seeing quality of the site.
An approach that met both of these objections had been developed at Kitt Peak for testing sites in the southern hemisphere. This was to use a pair of separated telescopes on a common mounting to form a pair of images of a single star: the seeing could then be assessed as the differential motion of the two - thus effectively eliminating the telescope shake which would be common to both images. To determine the diffraction-ring stability, our other main seeing-test method, we borrowed two Questar telescopes - relatively small instruments of about 5-inch aperture but with fine optical quality.
For monitoring meteorological conditions such as relative humidity, temperature, wind speed and direction, we were fortunate to have the assistance of Howard Ellis of the Mauna Loa Observatory, who provided instrumentation, while Jim Westphal, of Caltech, provided us with instruments to measure the infrared transmission of the overlying atmosphere via the relative attenuation of sunlight in two spectral bands.
Through the kindness of mainland colleagues, we were able to borrow much of the necessary site-testing equipment - Questar and double-beam telescopes in particular. As for staff, there was little demand in the Hawaii of those days for people with any technical training and the main supply was found among those few who washed up on the beach or were stranded on their way somewhere else. Such a one was Jim Harwood, who eagerly called me after reading an article about the proposed project in the Honolulu paper early in 1965. Once the money was in hand, I put him in charge of the Mauna Kea team, and together we started, in mid-July, learning how to use the equipment by setting it up in a picnic ground just above Hale Pohaku. A contemporary description of the results we obtained over the next six months or so is contained in a report presented to NASA to accompany our recommendation for Mauna Kea. Much of the following is taken directly from that (unpublished) report. 
At each site on Mauna Kea we set up (three-sided) canvas windscreens, 8-10 feet high, as crude shelters. These turned out to be entirely satisfactory for all but extreme conditions, but unfortunately extreme conditions were frequent, and we were soon forced to construct more robust shelters. We tested at four sites on Mauna Kea ranging from 12,000 feet up to Puu Poliahu (13,600 feet), where Arizona's telescope was located. On Haleakala we tested three sites around the summit ridge. Our first accommodation on Mauna Kea was in the rudimentary stone cabins that give the place its name, Hale Pohaku; soon afterwards, through the vigorous efforts of Bill Seymour of Hilo, we obtained a house trailer from government surplus and installed this in a fenced enclosure of the State Park Service. A second surplus trailer was hauled up the road to 12,000 feet, where it served as a much-needed refuge for the frozen observers as well as a storage unit.
As the Preliminary Report acknowledges, the staff of observers at both sites showed a remarkable level of commitment, and on Mauna Kea under conditions that were extremely rigorous - indeed hazardous to a degree that I shudder now to recall. We were all immensely exhilarated to be sharing in this adventure, and thrilled to be in at the start of what we were all convinced was an epochal time for the history of ground-based astronomy and for the University. While the Haleakala survey was relatively straightforward, I surely underestimated the difficulties of working on Mauna Kea - high, cold, dangerous, uncomfortable, and unutterably lonely as it was. The spirit that infused the group that worked there cannot be too highly praised. Their enthusiasm, initiative, and willingness to make do showed a pioneering spirit that presaged one that would characterize the Institute for years into the future.
After six months of testing, we concluded from our data that there was a distinct advantage to Mauna Kea in each of the primary determinants - seeing quality, freedom from cloud cover, and atmospheric transmission in the infrared. However, its high altitude remained a concern. We had cast a wide net seeking assistance from people within the military and academia who might give us some guidance on human performance at high altitude, but we could find little relevant to our situation. In the end, we decided that the ability of our observers to function at the summit under conditions far more stringent than those that would be asked of the telescope users was evidence enough that the altitude would not be a fatal hindrance - though they would certainly know they were not at sea level! This, of course, assumed that people working at the summit would follow the same precautions as our site-survey staff in eating and drinking moderately - and especially in acclimating for 24 hours at Hale Pohaku before observing at the summit, and this was instituted as standard practice. Those visitors for whom this precaution was not practicable, or who ignored it, could anticipate more or less severe consequences from the altitude - nausea, vomiting, crashing headaches were my own rewards for such visits. There was also the possibility of pulmonary edema - for signs of which potentially fatal condition we had to be constantly vigilant.
So, not without some trepidation, we recommended that the telescope be built on Mauna Kea and, in March 1966, NASA accepted this. I had a most cordial letter from Kuiper (in response to my telegram letting him know that Mauna Kea had been selected) saying that it was a great decision for astronomy and I guess it was, at that.
The 84-inch proposal stated that the University of Hawaii would be responsible for constructing a surfaced road and a power line, as well as supporting buildings. A start on these was, however, delayed while we decided whether to build the road up the western (Kona) side or to follow the existing southern route. A decision between these was too-long delayed by my arguing for the Kona access, and in the meantime Governor Burns (who also favored that route) was persuaded that a cable car should be studied – an option that I resolutely opposed. These issues contributed to delaying the initiation of the infrastructure improvements for several years, and it was not until nearly the end of my tenure there that final approval was given to a power line and road. The mid-level facility – another essential element in the infrastructure - was handled rather more expeditiously and I had the pleasure of participating in the design and subsequent dedication of this fine set of buildings prior to leaving the islands.
1 "Preliminary Report on a Site Survey for an 84" Telescope' by J. T. Jefferies and J.B. Zirker. Jack had handled the survey while I was away for a few months because of a prior commitment.
2 A decade or so later the whole question of acclimatization was studied extensively by a physician attached to the UKIRT project. His conclusions bore out our early anticipation.
Mauna Kea Observatories website