572 International Organization 24 22 20 18- 16- 14 12- 10 6 2 0 0.0-0.30.3-0.60.6-0.90.9-1.21.2-1.51.5-1.81.8-2.12.1-2.42.4+ Percentage of gross domestic product spent on R&D FIGURE 1.Research and development(R&D)spending at the time of science policy adoption (Pakistan)to a high of more than $9,000/year(Denmark)at the time these bureaucracies were created. (4)Defense spending as a percentage of GNP ranged from 0.7 percent (Mexico,Sri Lanka)to more than 10 percent at the time of adoption(France, Iraq,Jordan,and the Soviet Union). The range of variation on the defense variable is more than a factor of ten; the range of variation on all of the other variables is a factor of one hundred or more.Ranges of variation this large do not readily suggest any causal connection between sufficient state conditions and the adoption of science bureaucracies. Similarly,Figures 1-4 provide little support for the necessary condition hypothesis,that is,that there is some minimum threshold value of these variables that triggers demand for the bureaucracy.If such a value existed,we should see very few (or no)adoptions at the low end of the value range for one or more of these variables;all values would be spread across the upper end of the range at or above the necessary condition level. Again,the far-flung distribution of values revealed in these figures and elaborated in the text above does not readily support this proposition.Rather than clustering at the upper end of the value ranges,there seems to be a concentration of values at the low end,particularly on the science variables. This is clearly not a bunching that would support the existence of a necessary572 International Organization 24—i 22- 20 — 18 — 16 — 14— 6 — 4 — 0.0-0.3 0.3-0.6 0.6-0.9 0.9-1.2 1.2-1.5 1.5-1.8 1.8-2.1 2.1-2.4 2.4+ Percentage of gross domestic product spent on R&D FIGURE l. Research and development (R&D) spending at the time of science policy adoption (Pakistan) to a high of more than $9,000/year (Denmark) at the time these bureaucracies were created. (4) Defense spending as a percentage of GNP ranged from 0.7 percent (Mexico, Sri Lanka) to more than 10 percent at the time of adoption (France, Iraq, Jordan, and the Soviet Union). The range of variation on the defense variable is more than a factor of ten; the range of variation on all of the other variables is a factor of one hundred or more. Ranges of variation this large do not readily suggest any causal connection between sufficient state conditions and the adoption of science bureaucracies. Similarly, Figures 1-4 provide little support for the necessary condition hypothesis, that is, that there is some minimum threshold value of these variables that triggers demand for the bureaucracy. If such a value existed, we should see very few (or no) adoptions at the low end of the value range for one or more of these variables; all values would be spread across the upper end of the range at or above the necessary condition level. Again, the far-flung distribution of values revealed in these figures and elaborated in the text above does not readily support this proposition. Rather than clustering at the upper end of the value ranges, there seems to be a concentration of values at the low end, particularly on the science variables. This is clearly not a bunching that would support the existence of a necessary