lap and the desire to avoid resolving disputed facts apparently were the sole motivations behind the sentence imposed on Willard.
The judge, after eliminating any need to choose between the ranges, did belatedly state that resolving the dispute “wouldn’t really make any difference one way or another I think' under the facts of this case.” But, for several reasons, we conclude that the judge did not fulfill his responsibility to satisfy himself that, if there had been no dispute about the applicable range, the sentence imposed would have been the same under either of the ranges presented to him.
First, the equivocality of the judge’s denial of “any difference” contrasts sharply with earlier unequivocal statements that he imposed an 87-month sentence primarily because the existence of the overlap allowed him to avoid resolving the dispute. The sentencing record contains ample information to conclude that the overlap itself was the motivation for the sentence, not the fact that the sentence would be the same under either of the ranges in the absence of any dispute.
Second, the judge did not express any appreciation for the factual differences between the Government’s and the defendant’s sentencing positions. The Government’s position was that Willard should be held responsible, as a coconspirator, for the illegal distribution of drugs with the heroin equivalency of over 8 kilograms. Successfully proving that contention, or responsibility for any amount between 3 and 9.9 kilograms, would have resulted, after various calculations, in an applicable sentencing range of 87 to 108 months imprisonment.
Willard, however, contended that the extent of the entire conspiracy was not known or reasonably foreseeable by him. Apparently, he tried to show that he should be held accountable only for the distribution of narcotics with the heroin equivalency of about 1 kilogram. If Willard were successful in limiting his responsibility, as a coconspirator, to 1 kilogram, or indeed any amount less than 3 kilograms, the applicable sentencing range, after adjustments, would have' been 70 to 87 months. Given these facts, the culpability of Willard would seem significantly to depend on whether the Government or Willard is right as to a basic fact that has never been resolved. It is by no means apparent why Willard's sentence would be the same regardless of who is right in the dispute. Not only did the sentencing judge fail to articulate why he considered 87 months to be appropriate irrespective of the applicable range, he failed even to mention his awareness of the factual differences, as discussed above, between the Government’s and the defendant’s positions.
Furthermore, the Government agreed, in the plea bargain, to “recommend that [Willard] be sentenced to the low end of his final guideline range.” Willard’s intention to rely on the lower end of the applicable range was a point maintained by him and a point with which the Government fully concurred from the time the plea bargain was entered. While the sentencing judge is not bound by the Government’s recommendation, the judge’s lack of mention of any reason why he was going to ignore it leads further to the conclusion that Willard might have received a sentence of less than 87 months if the lower range had unquestionably applied. The language
of United States v. Bermingham, 855 F.2d *925* (2d Cir.1988), is especially pertinent:
Situations may arise ... where a sentencing judge determines that the appropriate sentence is whatever number of months are at or near the bottom of the applicable guideline range. In such situations, it obviously matters which guideline range is applicable, and a judge entertaining such a view could not conscientiously be satisfied that the same sentence would have been imposed simply because it happened to fall at or near the high end of a lower, overlapping guideline range. In such circumstances, any dispute as to the applicable guideline range must be adjudicated in the trial court, and that adjudica