[1] Defendant contends that he was denied effective assistance of counsel. He argues that his appointed counsel prepared inadequately for trial and conducted a preparatory interview with his alibi witness in the presence of the prosecutor. These omissions and mistakes, he says, caused him substantial prejudice at trial.
In assessing defendant’s claim, we must determine, considering all the circumstances, whether his attorney rendered him “reasonably effective assistance.” Strickland v. Washington, — U.S. —, 80 L.Ed. 2d 674, 693-94, 104 S.Ct. 2052, 2064-65 (1984). If defendant’s attorney did not render such assistance, then we must determine whether prejudice resulted, ie., whether “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Washington, 80 L.Ed. at 698, 104 S.Ct. at 2068. Our analysis of “reasonably effective assistance” and prejudice reduces essentially to whether defendant’s trial was fundamentally fair. Our purpose is to ascertain whether counsel’s performance so handicapped the presentation of defendant’s case and so impaired the functioning of the adversarial process that we cannot say with confidence that the trial produced a just result.
Defense counsel’s handling of his client’s case certainly left much to be desired. Counsel apparently failed to interview witnesses for defendant prior to trial, or to take timely steps to assure they would be present at trial. Two subpoenas were drawn up on the day of trial and another was not delivered properly to the Sheriffs office, so that it was never served. During the defendant’s trial, the trial judge sent deputies out to locate witnesses and summon them to court.
When the witnesses eventually arrived, defense counsel attempted to ascertain how they would testify. He apparently did *723so within the hearing of the prosecutor. When one of the witnesses took the stand, the prosecutor sought to impeach her by referring to statements he had overheard a few minutes earlier.
Defense counsel did effectively cross-examine the State’s witnesses, and also effectively examined his client and a police officer brought to the stand for the defense. Yet, we find that counsel’s failure to interview potential witnesses prior to trial, his failure to take timely steps to bring them to court, and his interviewing of defense witnesses in court, within the hearing of the prosecutor, reflected not sound trial strategy, but serious neglect of his client’s interests. In light of these circumstances, we cannot say that defendant received reasonably effective assistance of counsel.
Our analysis of whether defendant was rendered reasonably effective assistance virtually answers the second question, of whether defendant was prejudiced. The circumstances of this case indicate that counsel so compromised defendant’s case that we cannot with confidence say that the adversary process produced a just result. Had counsel not made such unprofessional errors, a reasonable probability exists that the jury’s verdict would have been different.
Our conclusion that there is a reasonable probability that a different result would have been reached is supported by the difficulty the jury had in reaching its verdict and by the inconsistency of the verdict itself, a split verdict, with the facts of the case as presented in the evidence.
[2] Defendant contends also that the trial judge’s statements to the jury during deliberation, and especially his inquiry as to the jury’s numerical division, were coercive and violative of the right to trial by jury guaranteed by Article I, § 24 of the North Carolina Constitution. We agree.
In State v. Yarborough, 64 N.C. App. 500, 307 S.E. 2d 794 (1983), this Court set out the standard for review in cases where the trial judge has inquired into a jury’s numerical division. In that case, we declined to adopt a per se rule allowing a new trial each time such an inquiry is made, but observed that such an inquiry can, in certain circumstances, be useful. Yarborough, 64 N.C. App. at 502, 307 S.E. 2d at 795. To determine if it has caused *724undue pressure on the jury, however, we must “examine the trial judge’s inquiry in context of the totality of the circumstances.” Id.
In the case at bar, the jury had deliberated the afternoon of 19 October and the morning of 20 October without reaching a verdict. At 12:20 p.m. on 20 October, the jury returned to the courtroom, and a dialogue ensued between the foreman and judge. The judge said to the foreman, “Well, I haven’t heard a knock on the door.” The foreman replied, “That’s right.” The judge then said, “What about it, Mr. Foreman?” The foreman replied, “Your Honor, we’re trying to be fair. All that’s involved, it’s a very unique problem.”
The judge then asked the foreman to give him the division of the jury, without any indication which way the vote was going. The foreman reported votes on the two charges of 9-3, 8-4, and 10-2, 5-7. The foreman stated that the jury was “at a standstill,” and that he felt that the jury “would have a hard time being unanimous under the circumstances.” The trial court stated:
I don’t want to lean on you too heavy. I don’t want to make you do anything that would go against any of your consciences but if you think you can reach a verdict, we will let you try but if you don’t think you can that’s another problem. You’ve been out there now since yesterday about 2:30 until 5 yesterday and you’ve been out this morning from 9:30 to 12:30. Has there been any change in the vote in the last hour?
Foreman: Yes, your Honor.
The Court: Maybe you’re making a little progress. Do you want to come back at 2 o’clock and try awhile longer? What do you all say? Let’s make one more effort after 2 o’clock and if you feel like you can’t reach unanimous verdicts in either one or both of these cases, then let us know, okay? We will take a recess now until 2 o’clock. Don’t talk about the case during the recess and come back at 2 and we will go on a little further. Okay?
Thus, although the numerical split was fairly wide, and the foreman stated that the jury would have a hard time being unanimous, the court still desired that the jury should keep trying to reach unanimous verdicts. At this point in the deliberations, the *725judge could have, and perhaps should have, declared a mistrial. Admittedly, whether to go on was a matter within his discretion. In the circumstances of this case, however, if no mistrial was declared, the better practice would have been to stress more clearly that each juror must decide for himself and not surrender his convictions for the mere purpose of returning a verdict. Indeed, the best practice would have been simply to repeat in toto the instructions of G.S. 15A-1235(b). The judge’s failure to do this, the fact that the verdicts were eventually reached only twenty-four minutes after the jury returned to deliberate (while the jury had deliberated up to that point for five hours), and the fact that the split verdict returned was inconsistent with the facts of the case, suggest that the judge’s remarks did influence certain members of the jury to agree, against their consciences, to unanimous verdicts. In the context of the totality of the circumstances, we find that the trial judge’s remarks so influenced the jury that a new trial is in order.
New trial.
Judges Wells and Hill concur.