In my view it is very much to their credit that NSF program managers have always been wise and pragmatic about this.

The four Indian Ocean legs, from December 1977 to April 1978, were thus a much smaller effort. Taro Takahashi found that the CO₂ problem grew yet worse, so that the measured pCO₂, and that calculated from the measured alkalinity and total CO₂, differed by well over 30 ppm (parts per million). The urge to tackle directly the growing fossil fuel CO₂ problem was now very strong, and this stood in the way. The elegant CO₂ model by Hans Oeschger and colleagues in Switzerland had appeared two years earlier and had stimulated renewed interest.

New interests were also arising; the "particle reactive tracers" such as thorium and ²¹⁰Pb were proving more tractable than anyone had thought—not for the original problem of the abyssal circulation rate, but for insights into how the ocean biogeochemical cycle worked. The program was evolving, and important breakthroughs in trace-metal geochemistry, organic geochemistry, and observing the rain of particles to the seafloor were occurring.

The general release of data from the shipboard program was keenly sought, but those close to the measurements were always aware that things could be improved. More problematic still were the results from the shore-based laboratories. These were closely held by the principal investigators (PIs) so as to maximize their advantage in publication; yet for such a conspicuous program there was widespread desire for full disclosure. The data release problem is commonly dealt with by NSF today, but it was the pattern that was created during the GEOSECS era that laid the rules. Pressure, official (the purse string) and peer, was brought to bear on PIs, and the results emerged.

The GEOSECS period resulted in all manner of fundamental insights. I wrote a paper on modifying the equation of state, thus eliminating an ambiguity in connecting the interocean abyssal flow (Brewer and Bradshaw, 1975). Wally Broecker carried on an amazing and sustained attack on the use of the fundamental nutrient relationships to decode water masses, mixing, and chemistry (beginning with Broecker, 1974; see Broecker and Peng, 1982, for a masterful analysis). And eventually the long haul of collecting, stripping, and measuring the radiocarbon signal was completed. Stuiver et al. (1983) published a remarkably simple and elegant paper compiling the ¹⁴C results (Figure 2). They reported that "the mean replacement times for the Pacific, Indian, and Atlantic ocean deep waters (more than 1500 meters deep) [are] approximately 510, 250, and 275 years respectively. The deep waters of the entire world ocean are replaced on average every 500 years." These ages were much shorter than first expected: the promise of a radiocarbon solution to the "age" problem was fulfilled; the task of shedding more light on geophysical fluid dynamics by the tracer approach proved to be far more complex. Of course such a paper was principally a necessary and welcome formality. Thanks to the wise NSF data release policy, the GEOSECS results had already been in use around the world, for all kinds of innovative uses, for many years.

Figure 2 The ?¹⁴C values of the cores of North Atlantic, Pacific, and Indian Ocean deep waters. The oldest waters are encountered near 40°N in the Pacific Ocean. Reprinted from Stuiver et al. (1983) with permission from the American Association for the Advancement of Science.

In 1978, John Steele called me down to his office at the Woods Hole Oceanographic Institution (WHOI), to meet with John Ryther and Hank Stommel. He wanted to see some fresh starts, and he was concerned about the ending of GEOSECS. He particularly wished to see WHOI tackle the CO₂ problem in some way. Hank referred him to a short NAS report, written with Jules Charney, on the anticipated thermal changes; I volunteered to look at the GEOSECS data to find the oceanic chemical signal. Since I had served as co-chief scientist on GEOSECS Atlantic Leg 6, I simply went to those data and wrote a provocative paper on the procedure for detecting the fossil fuel CO₂ signal above the very large natural background.

At the same time, Gote Ostlund in Miami was fretting about the lack of a GEOSECS follow-on. The classic dilemma with large programs is the problem of continuity versus innovation; a superb observing system bad been created and refined, and a talented team of people, particularly the Operations Group under Arnold Bainbridge at Scripps, existed. I had heard that Gore was to hold a meeting, with Department of Energy support, to discuss this and I called him to ask if I could attend. I gave the fossil fuel CO₂ paper and pointed out that, due to the early GEOSECS technical problems, we had no Atlantic data north of 20°N in the critical deep water formation regions. Others pointed out the new information from the chemical tracers in the region, and we