GUSTAV KRATZ AND GRANGE MUTUAL CASUALTY COMPANY, A FOREIGN CORPORATION, APPELLANTS,

v.

GARRY V. NEWSOM, APPELLEE

MANN, Judge.

We are split 2-2-1. There would ordinarily be no need for a third opinion on the same question, but the prior opinions 1 of this court in two cases like this, in which the trial judge has declined to sever an insurance company from trial, are in conflict. It is then apparent that the result in such a case is that its outcome on appeal depends on the random choice of the panel before which it is heard. My view is that these three cases should be affirmed, but that in the future trial judges should grant severance unless some reason for the insurer’s participation at trial is shown by the plaintiff, and I would reverse in cases tried hereafter. We are all so uncertain about the meaning of our Supreme Court’s opinion in Beta Eta 2 that we certify the questions involved here for authoritative resolution.

To summarize, Chief Judge Pierce and Judge McNulty would reverse. In their view the older cases to the effect that the existence of insurance may unfairly prejudice a trial have not been overruled. As our Supreme Court said in Beta Eta, “The existence or amount of insurance coverage has no bearing on the issues of liability and damages, and such evidence should not be considered by the jury.” The plaintiff having shown no reason for the insurer’s open involvement at trial, the motion to eliminate the insurer from the trial of liability and damages should have been granted and failure to do so was reversible error, in their view.

Judges Hobson and Liles would affirm because the appellant has not demonstrated any abuse of judicial discretion, as *540Judge Hobson explained in Hartford Accident & Indemnity Co v. Myers 3. Since the insurer is, since Shingleton v. Bussey,4 a proper party and there is no absolute right to severance, there was no reversible error.

The outcome of this case is the same as if the case were heard en banc because of my position, which, simply stated, is that I agree with Judges Liles and Hobson on this case and with Judges Pierce and McNulty on the next one if tried after the date of this opinion. What may seem on the surface to be a wishy-washy attitude is underlain by what I think is sound law.

At the root of the problem is a misconception about “judicial discretion.” In Beta Eta our Supreme Court led many trial judges plausibly to believe that the grant of severance was a question within the judge’s “discretion.” I think this word grossly misused. In its dictionary sense, discretion implies that the person exercising the discretion may do as he pleases without objection from anyone. Whether one orders beef or chicken in a restaurant is a matter of discretion. When we tack the adjective “judicial” to the word we invite misunderstanding. Properly used, “judicial discretion” implies a judgment which is, for practical purposes, unreviewable. It involves the exercise of judgment by a trial judge whose presumptive correctness is heightened by a view of the situation which an appellate judge cannot have, or which -a written record cannot reflect. It involves a sense of situation which may be wrong, and may seem wrong to an appellate judge, but as to which the likelihood of misjudgment is so great as to make it more just to allow the matter to stand.

When we come to the phrase “abuse of discretion,” we have another concept in play. We might refer either to a matter normally unreviewable where the departure is so egregious as to warrant appellate intervention. Or we might mean simply that the trial judge has applied a wrong principle of law. The phrase comes so readily to the mind of a judge burdened with more opinions to write than he should have that the temptation is great to use it.

I think that a question susceptible to the application of legal principle is not a matter of “discretion.” Consequently, I think trial judges should routinely sever insurance companies from trial because the existence of insurance is irrelevant and because the same question should get the same answer in a court of law. Our Supreme Court has in past years thought the mention of insurance in such a case as this a reversible lapse, as Judges Mc-Nulty5 and Pierce6 have explained. It has also expressed the view that even after Shingleton v. Bussey the existence of insurance may be prejudicial. Consequently, the better practice would be to sever unless there is some reason peculiar to the case which demands the presence of the insurance company at that stage.

There is virtue in principle. It is that litigants can know that the rule of law, not the practice of a particular judge, will govern the conduct of legal proceedings to the maximum extent possible. I think this matter is susceptible to the formulation of principle, and would hereafter vote to reverse any judgment where the error assigned is failure to sever the insurance company from a trial in which no cause for its retention is shown. As matters now stand we have circuits in which some judges do, and some do not, sever. It is conceivable that future juries may come to think the presence of the insurance company the standard thing and a heavily insured defendant in court with*541out the visible presence of his insurer may gain sympathy. That is just as bad as a plaintiff’s gaining it where the idea is conveyed that the damages come out of a deeper pocket. From the standpoint of uniform and principled practice, it is better that the fact of insurance be excluded from the trial.

The question logically arises why I would not reverse in this case. That question too has a logical answer. Any appellant has the burden of showing that the trial judge has committed error. I think it is presumptuous of this court to tell this trial judge that he erred in doing what the Supreme Court said he had the “discretion” to do. There is no showing on this record that the presence of the insurer at trial actually influenced the jury. No claim is made except that in years past the mention of insurance was error, and that Shingleton and Beta Eta have not changed this. Probably this is a sound position. The Florida Bar’s Practice Manual No. 13, Evidence in Florida, cites the old rule in § 1.14, and goes on to say in § 1.18:

“Joinder of a liability insurer as a party defendant will not necessarily result in knowledge by the jury that the tortfeas- or has liability insurance because the court, in its discretion, may order separate trials of separate issues such as denial of coverage by the insurance company or a claim of bad faith against the insurance company in settlement negotiations. Beta Eta House Corp., Inc. of Tallahassee v. Gregory, 237 So.2d 163 (Fla.1970); Shingleton v. Bussey, 223 So.2d 713 (Fla.1969).

“Although the Shingleton and Beta Eta cases have made a major procedural change in the law as it pertains to parties and trials, they are consistent with prior case law holding that evidence of insurance coverage is not relevant to the issues of liability or damages.”

In a case like this where the appellant can demonstrate no prejudice flowing from the trial judge’s failure to sever, I would affirm. I have no hesitation in announcing a different view for the next case. The law is full of instances in which a judge-made rule, relied on by trial judges, is altered for the future by the candid recognition that the rule ought to be different from now on. Rather than elaborate this view, I refer the reader to more material than he will want to read on the subject.7

There is still the possibility, if other cases arise before a definitive answer from the Supreme Court, of results in this court varying according to the composition of the panel. In such a case we might entertain a petition for rehearing en banc and certainly would protect litigants against losing the case by virtue of random assignment of judges.

Affirmed.

HOBSON, J., concurs in the result for reasons expressed in Hartford Accident & Indemnity Co. v. Myers, Fla.App.2d 1971, 247 So.2d 83.

PIERCE, C. J., dissents for reasons expressed in his dissent in Hartford Accident & Indemnity Co. v. Myers, supra.