OPINION

Martinez and Sidle, who was charged and tried under the name Stephens, were convicted under two counts of an indictment charging them and one Marie Dab-ney, jointly, with importing 117 grams of cocaine, in violation of 21 U.S.C. §§ 952 (a) and 960(a)(1), and with possession of the same quantity with intent to distribute it, in violation of 21 U.S.C. § 841 (a)(1). Each received concurrent sentences on the two counts. In addition, Sidle was convicted under two counts charging him alone with importing a “measurable quantity” of cocaine, and with possession of the same quantity, in violation of 21 U.S.C. § 844 (a). He received concurrent sentences under these counts and these sentences were less than, but made concurrent with the sentences under the two counts involving the 117 grams. Dabney, being a fugitive, was not tried.

I. Facts

Appellant Martinez drove a Lincoln Continental sedan across the border from Mexico to Nogales, Arizona. In the front passenger seat next to him was one Julia Davis. In the back seat were appellant Sidle on the left behind Martinez, and Dabney on the right behind Davis. The car was searched at the border customs and immigration station. In a facial tissue box on and in the center of the dashboard agents found a “zi-plock” plastic bag containing 117 grains of cocaine. In Sidle’s pocket they found a small gilt or gold cocaine spoon on a gold or gilt chain. The spoon, which is very small, appeared to have a residue on it. The car was leased to Dabney; it was not the usual rental agency car. Davis had $1,000 in cash on her person. When asked about the spoon, Sidle said that it was Dabney’s. Dabney said that the spoon was hers, that she did not want to see the other three go to jail and was willing to take the blame, and that she loved cocaine. Neither Martinez nor Sidle admitted knowledge that cocaine was in the car.

No fingerprints were taken from the plastic bag containing the cocaine or from the facial tissue box. A government chemist testified that “it appeared that there was something on the bowl of the spoon” but that it was not possible to “make out what it was” until it was placed under a microscope. The chemist also testified that “an inexperienced person glancing briefly at something like that, not knowing what to expect, probably would not readily distinguish there was something there. If someone who is familiar with things like this, for whatever they are more or less generally thought to be used for, inexperienced [sic — experienced?] in working with things like that, and were suspected prohibited or controlled materials, there was enough there that an experienced person would in a brief examination be able to detect that there were things there.” In order to determine what the residue on the spoon was, the chemist rinsed it off the spoon and employed chemical tests. He did not determine the amount of material on the spoon. He testified that there was less than one-tenth of a gram (less than V28oth of an ounce), but did not say how much less.

Sidle did not take the stand, but Martinez did. Martinez’s story was that Sidle, whom he had known for about three months, and the two women, whom he had never met, came to his house and asked him to go to Mexico as their chauffeur and translator. Martinez speaks fluent Spanish. The object of the trip was to “party” for a couple of weeks, with Dabney and Davis “treating.” On cross-examination, Martinez said that for the entire “few days” while the four were in Mexico until the day they were arrested, they were together. However, he claimed that on that day the four of them were in a Canal Street bar drinking when the two women asked for the car keys and left. He said that they left at about noon and returned at about one or one-thirty (p.m.), and asked Martinez and Sidle to take them back to the United States. According to Martinez, the four then left the bar and “rode around town for a little bit more and then from there we went out to the border.” Asked at what time they arrived at the border, Martinez replied that “[i]t was pretty dark. It was getting dark. I guess we got to the border about 4:00 or 3:00 o’clock.” Martinez also admitted that he had once tried cocaine “about three years ago” but said that he hadn’t liked it and had returned to drinking. However, he denied familiarity with the appearance of cocaine. On cross-examination, he admitted that he knew that cocaine “looks like sugar” and is a “white powder.”

II. Martinez’s Appeal (No. 74 — 2825)

A. Denial of motion for acquittal when the government rested.

Martinez argues that it was error to deny his motion for a directed verdict of acquittal following the close of the government’s case, because there was no¿evidence of his knowledge that the cocaine was in the car. However, Martinez elected to proceed with the presentation of proof in his own behalf and thereby waived his right to challenge the denial of this motion on appeal. United States v. Figueroa-Paz, 9 Cir., 1972, 468 F.2d 1055 , 1058 — 59; United States v. Lewis, 9 Cir., 1970, 426 F.2d 266, 267 ; Benchwick v. United States, 9 Cir., 1961, 297 F.2d 330, 335 .

B. The aiding and abetting instruction.

Martínez claims that it was error for the judge to give an aiding and abetting instruction because aiding and abetting was not specifically charged in the indictment. He is wrong. United States v. Roselli, 9 Cir., 1970, 432 F.2d 879, 895, n. 27 .

C. Sufficiency of the evidence.

Martinez argues that the evidence is insufficient to support the verdicts of conviction. The case is a thin one. We have found no decision of ours affirming a conviction on facts as weak as these. On the other hand, we have found none reversing on similar facts.

1. Inferring knowledge from the fact of driving the car.

Running through a number of our recent cases, many of them short per curia, are statements substantially like the following:

On its face this appears to be an easy. “bright line” rule of law, dispositive of Martinez’s argument. However, a review of our cases shows that the rule is based on the idea that a driver exercises dominion and control over his vehicle and its contents, from which dominion and control “knowing possession” of the contraband that it contains may be inferred. The facts of our cases fall into four categories: those in which the driver was the only occupant of the vehicle, those in which the driver was the owner of the vehicle, those in which the driver had borrowed the vehicle from someone not present in the car at the time of the arrest, and those in which, for some other reason, it could be said that the driver was in complete control of the vehicle. Sometimes, as might be expected, these categories overlap.

Here the only circumstance tending to indicate that Martinez controlled the car was the fact that he was driving. “Title” to the car, as lessee, was in Dabney’s name, not Martinez’s name, and Martinez, at 21, was significantly younger than the other occupants. Dabney was 33 and Sidle was 42. Davis, the fourth occupant, had a thousand dollars in cash, which apparently was what the group was subsisting on. No significant amount of cash was found on Martinez or Sidle, and Martinez was apparently unemployed. It is a dubious proposition to say that the driver has dominion and control of the car and its contents on these facts, and that therefore knowing possession can be inferred.

Most of the cases cited in the previous footnotes also involved concealment of contraband in secret compartments, locked trunks, beneath or behind seats, or in other more or less elaborate hiding places in the physical structure of the vehicle. Where such a vehicle is stopped and contains other passengers, it is reasonable to infer that it was the driver, when he had control of the car before the trip, who did the preparation necessary for the concealment. But where the contraband is found on the person of a passenger, or in a loose object like a tissue box, not obviously in anyone’s possession, no such preparation is necessary, and the inference of the driver’s knowing possession from his control of the car seems substantially less reasonable, especially, as here, when there are three other passengers in the car.

In Murray v. United States, 9 Cir., 1968, 403 F.2d 694 , two brothers had gone to Mexico in a borrowed car. Lonnie was driving, and at the border upon their return to the United States, heroin was found secreted on Johnnie’s person. The trip had been Lonnie’s idea and there was some indication that he had a stronger personality than Johnnie and might even dominate him. Nevertheless, we reversed Lonnie’s conviction on the ground that he was not in constructive possession of the contraband because “he had [not] personally shared custody of [the drug] or had such dominion and control over the drug as to give him the power of disposition.” Id. at 696 . The difference between Murray and the case at bar is that the contraband was not found on a passenger’s person, but in a loose object accessible to all the occupants but within the unambiguous dominion and control of none. Thus this case is like Murray in that the driver did not have sole dominion and control over the tissue box, and different in that the box was not obviously within the dominion and control of one of the passengers. We think that it is the similarity, rather than the difference, that is crucial. As we said in Delgado v. United States, 9 Cir., 1964, 327 F.2d 641 , a case in which marijuana was found in a bedroom nightstand of an unmarried couple living together, “[i]t is fundamental to our system of criminal law that guilt is individual. Here, that means that there must be sufficient evidence to support a finding, as to each defendant, that he or she had possession of the marijuana.” Id. at 642 . Here, as there, it is close to “pure speculation,” in the absence of additional evidence, that the driver Martinez knew of the contraband in the tissue box.

2. Inferring knowledge from a joint venture.

We have held that, regardless of which defendant is driving, where a foray into Mexico results from a joint design, knowledge of the presence of contraband in a contraband laden vehicle may be inferred as to all occupants. United States v. Campbell, 9 Cir., 507 F.2d 955 (1974); Eason v. United States, 9 Cir., 1960, 281 F.2d 818 . But the evidence of joint venture in these cases were significantly stronger than the evidence here. In Campbell all four occupants of the car had apparently been friends for some time. Additionally, the holding in that case was not based entirely on the joint venture theory, but also on the defendant’s “complete control” of the car and the fact that the defendant himself appeared nervous. Also, the defendant’s story about working for the Los Angeles Probation Department proved to be false when the customs inspector asked the defendant for identification. Similarly, in Eason both occupants of the vehicle had been good friends for two years, and both appeared nervous upon their arrival at the border. Also, the contraband in both Campbell and Eason was secreted in the car rather than being found on an occupant’s person or in a loose object in the car. And see United States v. Irion, 9 Cir., 1973, 482 F.2d 1240 ; United States v. Aranda, 9 Cir., 1972, 457 F.2d 761 , both involving stronger evidence of a joint venture.

These cases are difficult to reconcile with Murray, supra. In Murray we rejected the theory of joint venture-joint possession, even though it was not disputed that the brothers went to Mexico together, and that the trip was Lonnie’s idea. Is going to Mexico with good friends a joint venture, but going with one’s brother not a joint venture? The opinion in Murray, 403 F.2d at 696 , hints that the government must demonstrate some connection between the joint venture of traveling to Mexico and the narcotics,- but no such connection is apparent from the opinions in Eason and Campbell. We think that the only way that these cases can be reconciled is to say that for a conviction to be sustained the government must have shown something more than a joint venture which indicates that a particular defendant has knowledge of the contraband. Here there was nothing more than that Martinez was driving and that he was friendly to the customs agent.

In the case at bar, the only evidence of a joint venture came from the mouth of Martinez. He testified that he had known Sidle for about three months, but that he had never met the two women before the day they all left for Mexico. This case resembles United States v. Bonds, 9 Cir., 1970, 435 F.2d 164 . There the defendant was a passenger in a car rented and driven by a co-defendant, who was nervous at the border, and a search disclosed contraband hidden under the rear seat. In reversing, we noted that

Here the government presented no evidence that the trip was jointly planned. The only evidence of that was Martinez’s uncontradicted testimony that he had gone along with the group as a chauffeur and translator, and this story is supported by the facts that Dabney supplied the car and Davis had the cash. While the jury was not required to believe this story, it is the only evidence of a joint venture. We find it very difficult to believe that Martinez participated in the planning of the trip as an equal partici pant. In this respect, this case resembles United States v. Steward, supra, and United States v. Gardner, 9 Cir., 1973, 475 F.2d 1273 , where the convictions of chauffeurs were reversed.

3. Inferring knowledge from false statements.

There is, however, more evidence, albeit not much more, and it, too, comes from Martinez. Disbelief of a defendant’s own testimony may provide at least a partial basis for a jury’s conclusion that the opposite of the testimony is the truth. United States v. Chase, 9 Cir., 1974, 503 F.2d 571, 573 ; United States v. Ledesma, 9 Cir., 1974, 499 F.2d 36, 43 ; United States v. Hood, 9 Cir., 1974, 493 F.2d 677, 681 ; United States v. Cisneros, 9 Cir., 1971, 448 F.2d 298, 305-06 . The scope of this principle is not settled, see Chase, supra, at 574-75 (Wallace, J., concurring), but at the very least

But such disbelief can provide only partial support: there must also be “other objective evidence on the record which buttresses the fact finder’s drawing of the opposite inference,” id., n. 4, citing Cisneros, supra, 448 F.2d at 306, n. 10 .

Here Martinez’s story about driving around for four and one-half hours was nothing if not implausible, and in addition the jury had other reasons to doubt his credibility: he lied about his familiarity with the appearance of cocaine, and testified that he arrived at the border at three or four o’clock when it was getting dark, a statement that is not only self-contradictory, but was contradicted by every customs officer that testified. Martinez admitted that the four occupants were together in Nogales at all times during their visit except for the one to one and one-half hours when he and Sidle were left in the bar by the two women, who “went for a ride.” His claim that he had no knowledge of the cocaine thus depends on the assumption that during that one and one-half hours, the two women, who he knew did not speak Spanish and did not know their way around the streets of Nogales, went out and obtained the contraband. This too seems implausible. Moreover, Martinez had a strong motive to exculpate himself. It remains true, however, that there is little other objective evidence in the record to buttress a fact finder in drawing the opposite inference, as Chase, supra, requires.

4. Conclusions as to sufficiency of the evidence.

Under our decisions, the fact that Martinez was driving, although not sufficient, standing alone, to support a finding of knowledge in this case, is a fact that the jury could consider. Similarly, the evidence of a joint venture could also be considered. The jury could find that Martinez told the truth as to the fact of a joint venture, but not about its purpose. When his other false testimony is added, we think that the jury could infer from all of the foregoing, and beyond a reasonable doubt, that he knew that the cocaine was in the box that was in and on the dashboard.

D. The instruction about the inference of knowledge.

The judge told the jury that if it found that either defendant did not know that the cocaine was in the car, the jury must acquit him. It also, over objection, gave the following instruction:

Counsel’s primary objection was that the court did not tell the jury that it need not draw the inference.

We think that, in a case such as this one, where the driving of the car is not, as we have explained, sufficient, standing alone, to support the inference of knowledge, the instruction was both erroneous and prejudicial. This is partly for the reason stated by counsel, but also because it permitted the jury to draw the inference solely from the fact of driving the car. It is true that the second sentence refers, in passing, to “all the other evidence in the case,” but the first sentence does not.

There will be many cases in which the instruction would be . perfectly proper. Here, however, we conclude that it improperly emphasized the driving of the car. An instruction, in this case, referring to the fact of driving of the car, if given at all, should make it clear that that fact, standing alone, is not enough to support an inference of knowledge, but that it is one fact which, together with the evidence of joint venture and Martinez’s own testimony, if the jury finds that he lied, can be considered by the jury in deciding whether or not he had the requisite knowledge. We also conclude that, where the case is as close as this one is, the jury, if it is told that it may draw the inference from certain evidence, should also be told that it is not required to draw the inference.

We conclude that Martinez should have a new trial. The instruction given could well have tipped the scales on the side of conviction.

III. Sidle’s Appeal (No. 74-2826)

Sidle makes several arguments.

A. Sufficiency of the evidence.

Sidle argues that his motion for acquittal should have been granted because the evidence against him was insufficient.

1. Counts I and II.

As to Counts I and II, which relate to the 117 grams found in the facial tissue box, the evidence against him, as it relates to his knowledge, is in one respect weaker than the evidence against Martinez. Sidle was neither the driver nor the owner of the car. In other respects, however, the evidence against him is stronger. The jury could find that he was the instigator of the joint venture. Moreover, he had the cocaine spoon, with its residue of cocaine, on his person. From this, plus Dabney’s statement that the spoon and the 117 grams were hers, a jury could infer that he knew about the 117 grams, especially when it also appears that he was the instigator of the junket. We conclude that the evidence against him was sufficient under counts I and II. See Bettis v. United States, 9 Cir., 1969, 408 F.2d 563, 568 ; United States v. Hood, supra; United States v. Ramos, 9 Cir., 1973, 476 F.2d 624 ; Deck v. United States, 9 Cir., 1968, 395 F.2d 89 .

2. Counts V and VI.

These counts charge importation and possession of a “measurable quantity” of cocaine, i. e., the residue on the spoon. The government failed to prove what it charged. The chemist’s testimony shows that the quantity of cocaine was not measured or measurable. All that the chemist could say was that there was identifiable cocaine. He did not know the quantity of the residue, much less the quantity of cocaine in the residue. We doubt that the charge would ever have been presented were it not for the cocaine in the tissue box.

B. The court’s questions to witnesses and comments.

We have examined the instances to which Sidle calls our attention. None merits discussion. The error, if any, was so minor as to be of no importance. The court’s conduct of the trial was exemplary.

C. The prosecutor’s closing argument.

In his closing argument, the prosecutor, in discussing the “knowledge” issue, said:

At this point the court sustained Martinez’s counsel’s objection, and gave the following cautionary instruction:

The crucial issue in the case against Sidle, with respect to both the cocaine spoon and the tissue box counts, was his knowledge — (a) that the spoon was a cocaine spoon; (b) that there was a cocaine residue on the spoon; and (c) that the 117 grams of cocaine were in the tissue box. Prior involvement could well persuade a jury to draw an inference of knowledge as to each of these matters. There was no evidence of such involvement, and Martinez testified that he had never known Sidle to use drugs during their three-month acquaintance. The prosecutor’s argument, directly before the case was submitted to the jury, told the jury that the government had evidence that Sidle had had prior experience. This was highly improper. Sidle’s possession of the cocaine spoon and residue permitted the jury to infer knowledge of both the residue and the gross quantity of cocaine, but the jury was not required to draw these inferences. Given the closeness of the issue, the prosecutor’s comment was extremely prejudicial. We do not think that the court’s prompt cautionary instruction could “unring the bell.” We have difficulty in viewing the argument as other than deliberate.

The government argues that the argument was permissible as a retaliatory comment aimed at defense counsel’s argument that if the government had any evidence of prior drug involvement it would have been presented. The defense argument may have been ill advised, but it would not justify the improper argument in retaliation.

The case against Sidle was thin. We conclude that the prosecutor’s argument, under the circumstances of this case, requires a new trial. See: United States v. Benson, 3 Cir., 1973, 487 F.2d 978, 981 ; United States v. LeFevre, 3 Cir., 1973, 483 F.2d 477, 479, n. 1 ; United States v. Schartner, 3 Cir., 1970, 426 F.2d 470 .

We have considered other arguments presented by Sidle, but find them to be without merit.

In No. 74-2825, the judgment is reversed and the case is remanded for a new trial.

In No. 74 — 2826, the judgment is reversed and the case is remanded for a new trial on Counts I and II. Counts V and VI will be dismissed.

. United States v. Castillo-Burgos, 9 Cir., 1974, 501 F.2d 217 (secret metal compartment); United States v. Dixon, 9 Cir., 1972, 460 F.2d 309 (trunk and beneath rear seat); United States v. Gonzalez, 9 Cir., 1972, 456 F.2d 1067 ; United States v. Ascolani-Gonza-lez, 9 Cir., 1971, 449 F.2d 159 (hidden compartment); United States v. Teran, 9 Cir., 1970, 434 F.2d 605 (behind rear seat); Gaylor v. United States, 9 Cir., 1970, 426 F.2d 233 (above radio under dashboard); United States v. Simon, 9 Cir., 1970, 424 F.2d 1049 (taped to car springs). . United States v. Trujillo-Tirado, 9 Cir., 1971, 448 F.2d 1269 (in bolted down tire in trunk); Bettis v. United States, 9 Cir., 1969, 408 F.2d 563 (under rear seat and in trunk); United States v. Jackson, 9 Cir., 1969, 408 F.2d 306 (wife, a passenger, threw contraband out car window with assistance from husband, who executed evasive maneuver in response to police pursuit and tried to mislead officer when apprehended); Eason v. United States, 9 Cir., 1960, 281 F.2d 818 (unclear which occupant owned car, but both nervous, and substantial evidence of joint venture introduced). . United States v. Campbell, 9 Cir., 1974, 507 F.2d 955 (under dashboard; driver nervous, substantial evidence of joint venture with three other occupants, court found driver in “complete control of the car”); United States v. Zumpano, 9 Cir., 1970, 436 F.2d 535 (secret compartment). . United States v. Seidman, 9 Cir., Sept. 18, 1974, 503 F.2d 1027 (driver, with female passenger drove camper, laden with fresh, green marijuana which reeked, 740 miles north to border; also, there was “other circumstantial evidence that appellant knew of his load”); United States v. Hood, 9 Cir., 1974, 493 F.2d 677 (husband driver, contraband found on floor of passenger seat where wife sitting (joint venture?), wife’s conviction affirmed as well, and court found husband “had control of the vehicle”); United States v. Zamora-Corona, 9 Cir., 1972, 465 F.2d 427 (“appellant had complete control over the car with little evidence of the passenger’s participation in its management and direction”); United States v. Guzman, 9 Cir., 1971, 446 F.2d 1137 (driver’s companion was her 12-year old daughter; contraband found in secret compartment). . The only case that we have found that does not readily fit in one of the four categories is United States v. Durgin, 9 Cir., 1971, 439 F.2d 685 . There 80 pounds of marijuana were found in a suitcase in the locked trunk of a car which the passenger had rented (see United States v. Fallang, 9 Cir., 1971, 439 F.2d 685 , a companion case). These cases are short per curia, and the facts given are so sketchy that it is impossible to tell whether there was substantial evidence of joint venture or whether for some other reason the driver was in control of the car or the trunk. . See also United States v. Steward, 2 Cir., 1971, 451 F.2d 1203, 1207 . “We cannot agree with the argument that Sands’ possession of the heroin can be inferred from the fact that he was the armed chauffeur of the car which brought Steward to the motel. Looking at the facts in the light most favorable to the government, there is nothing to establish that Sands exercised any dominion or control over the drugs. The cases relied upon by the government are distinguishable. They involve situations where the defendant had exclusive dominion and control over a vehicle in which the drugs were found hidden [citing Arellanes v. United States, 9 Cir., 1962, 302 F.2d 603 ] or where defendants had joint dominion and control over a vehicle wherein drugs were found secreted [citing Eason, supra]. The drugs were not found hidden in Sands’ car, they were in the possession of Steward.”