Hill, J.

Ammonia gas escaped from.aluminum tubing in a cold storage room (hereinafter referred to as Room 2) in a warehouse operated by the plaintiff, Oroville Cordell Fruit Growers, Incorporated (hereinafter called Oroville). This caused damage to the 16,608 boxes of apples then in Room 2. There is no question about Oroville being completely compensated for its loss. The controversy is over how that loss is to be apportioned between three insurance companies. Pacific Indemnity Insurance Company (hereinafter called Pacific), which had insured Oroville against loss or damage arising out of any accident, is concededly liable for the entire amount, unless there was an “explosion” within the purview of the extended coverage provision of the fire insurance policies issued by the defendants, Minneapolis Fire & Marine Insurance Company (hereinafter *119 called Minneapolis) and the New Hampshire Fire Insurance Company (hereinafter called New Hampshire). While there was no fire involved, each of the fire insurance policies also provided protection under the “Extended ■ Coverage Endorsement” for loss caused by “explosion.” 1 If there was an “explosion,” the extent of the liability of each of the insurance companies was likewise conceded to be: Pacific, $27,449.84 (50%), New Hampshire, ■ $21,959.87 (80% of 50%), and Minneapolis, $5,489.68 (20% of 50%).

The question presented is whether the manner of the escape of ammonia gas from the aluminum tubing constituted an explosion within the provisions of the Minneapolis and New Hampshire insurance policies.

Between 8 and 8:30 a.m. on the morning of December 28, 1961, the customary check of storage Room 2 revealed no trace of ammonia gas. Between 10:30 and 11 a.m. ammonia gas was detected in a room on the floor above Room 2; investigation disclosed that Room 2 was full of ammonia gas of such strength that a gas mask had to be used to enter the room. The supply of ammonia gas was cut off *120 at approximately 11 a.m. It took a considerable time, though blowers were employed, to get the fumes out of Room 2; and the investigation as to the point from which the gas had escaped was not made until the following morning when an elongated opening in the aluminum tubing (exhibit No. 1) was discovered.

The pressure of the ammonia gas normally was 20 to 30 pounds per square inch. Every 4 hours there was a defrost cycle, which lasted for a period of 15 minutes, and during this time the ammonia gas surged to pressures of up to 70 pounds per square inch. There had been such a defrost cycle from 10 a.m. to 10:15 a.m. on the morning of December 28.

There is no contention that there was any pressure in excess of 70 pounds per square inch in the tubing at the point where the gas escaped; nor is there any contention that this was an unexpected or unusual pressure. (In other parts of the refrigerating system where the pipes were carrying liquid ammonia, before its conversion to gas, the pressures would be 150 to 200 pounds per square inch.)

The aluminum tubing at the place where the ammonia gas escaped was kept completely covered with brine. Aluminum was not ordinarly used for such tubing, because brine has a corrosive effect on aluminum, 2 but the installation had been made in 1942 during World War II when more corrosive resistant materials (normally galvanized steel, galvanized cast iron, or admiralty metal 3 ) were not available. The tubing had been corroded until it was almost tissue thin in places, and none of the tubing was regarded as usable after December 28.

An expert witness testified that even the low pressure had become too strong for the tissue thin pipe and a hairline crack had developed through which the ammonia gas escaped. On contact with the brine, it formed a

*121 sodium hydroxide which is commonly called household lye, which is extremely corrosive and especially so to aluminum, and it proceeded over a period of time to eat away the lengthwise split into what is a horizontal or elongated hole . . . as shown in Plaintiffs Exhibit 1.

Asked as to the length of time required to make such a hole as shown in that exhibit, he replied, “Minutes. Aluminum is attacked very severely by caustic.”

After the parties had rested, the trial court, over the objections of all parties, made one of the experts its own witness for the purpose of determining which, if any, of the definitions of explosion proposed by the parties was applicable to what had transpired in the present case.

The following definitions were rejected by the expert as not being applicable to the circumstances here existing:

A sudden release of energy accompanied by noise, a change in volume, and a going away of materials from the source of the explosion.

A sudden and rapid combustion, causing violent expansion of the air, and accompanied by a report.

Rapid combustion, or other similar process, usually causing a loud report; a sudden and violent outbreak as of physical forces.

A bursting with violence and a loud noise, because of internal pressure, of a sudden bursting or breaking up or in pieces, from an internal or other force.

The definition which the trial court finally held to be applicable came from L. L. Olds Seed Co. v. Commercial Union Assur. Co., 179 F.2d 472 (7th Cir. 1950), and was:

“[A] sudden accidental, violent bursting, breaking, or expansion caused by an internal force or pressure which may be and is usually accompanied by some noise.” (p. 474)

Even this definition had to be tailored a little bit to fit, by omitting the word “violent.” The expert explained:

I think it fits except where the word “violent” could be replaced by “sudden.” I think the only thing lacking in that definition is that this wasn’t violent as far as mechanical force is concerned, but it is violent as far as result is concerned in that you are dealing with chemicals, not mechanical forces.

*122 Nevertheless, the trial court made a finding that:

[Tjhe pipe suddenly and violently ruptured, burst, and was blown open as a result of the internal pressure of the ammonia gas inside splitting the pipe. . . . [Finding No. 6]

And, upon that finding, the trial court based the judgments against New Hampshire ($21,959.87) and Minneapolis ($5,-489.68), these being the amounts for which liability was conceded if there was an “explosion” within the meaning of their policies.

We are convinced that the facts do not support the finding that the aluminum tubing was violently blown open. Even the,trial court’s own expert would not agree that “violent” was an appropriate description of what had happened.

The trial court may have been too much influenced by what seems to us the attenuated reasoning of some courts which take the position that though an explosion may vary in intensity — from a hydrogen bomb or an exploding boiler (with the classical explanation: the water was too low and the engineer too high), to the bursting of a matchhead or a cap in a toy cap pistol — it nonetheless meets the criterion of violence in an explosion. Their reasoning is that the insurer against loss caused by “explosion” should be liable for any loss that occurs in consequence of the escape of the contents of a pipe or other container, regardless of the degree of violence involved, unless falling within specific policy limitations. 4

We believe the better rule to be that the true meaning of “explosion,” in each particular case, must be settled not by any fixed standard, or accurate measurement, but by the common experience and notions of men in matters of that sort. That term in an insurance policy is to be construed in its popular sense, and as understood by ordinary men and. not by scientific men. Jersey Ins. Co. of New *123 York v. Heffron, 242 F.2d 136 (4th Cir. 1957); Commercial Union Fire Ins. Co. of New York v. Bank of Georgia, 197 F.2d 455 (5th Cir. 1952); Lever Bros. Co. v. Atlas Assur. Co., 131 F.2d 770 (7th Cir. 1942); Allen v. Insurance Co. of North America, 175 Pa. Super. 281, 104 A.2d 191 (1954); Morie v. New Jersey Mfrs. Indem. Ins. Co., 48 N.J. Super. 70, 137 A.2d 41 (1957); Couch on Insurance 2d, §42:56 (1962).

Let us assume, as testified by the expert on whom the trial court relied, that a hairline crack was formed in the tissue-thin aluminum tubing by reason of the pressure of the ammonia gas, and that the comparatively large aperture found in exhibit No. 1 was then made by the attack on the pipe by the sodium hydroxide, which was formed by the escaping ammonia gas coming into contact with the brine. There yet remains the question of whether there was sufficient violence involved in the creation of the hairline crack under the test which we have approved. Not even the expert suggested that there was “violence” in the usual connotation of that term. There was a violent chemical reaction when the ammonia gas and the brine came into contact, but that was not an “explosion.” ■

The trial court was admittedly groping to find a definition of “explosion” which would fit the occurrence, and, it seems to us, confused the violent effects caused by the escape of gas with the cracking of the pipe which permitted its escape.

The question of whether there was an “explosion.” should have been considered, not from the standpoint of. the expert, but from the standpoint of ordinary,men.

There is a further facet of the case,, which the trial court does not discuss and which, it seems to us, must be considered, and that is whether the brine which had corroded the aluminum tubing to tissue thinness did not also .finish the job and cause the opening in the pipe rather than the pressure of the gas from within. :

The theory that the ammonia gas may have escaped because the pipe was corroded through, finds strong support in our own case of Prentice Packing & Storage Co. v. United Pac. Ins. Co., 5 Wn.2d 144, 106 P.2d 314 (1940); it was not *124 an “explosion” case, but it did involve a claim on an insurance policy for damage caused by escaping ammonia gas. The policy covered loss occasioned by sudden and accidental tearing of the refrigerative unit caused by the pressure of the refrigerant. The testimony established that the pressure of the refrigerant could have caused the rupture of the pipe only if the pipe were worn to a thinness of approximately one-ten thousandth of an inch. A rupture did occur. However, the court stated that it did not necessarily follow that the pipe had worn to such a thinness. A verdict, allowing a recovery, was reversed on the ground that the verdict was mere conjecture and not based upon the evidence. According to the court (p. 162):

It might well be argued, as appellant contends, that, under that theory, the proximate cause of the rupture was not internal pressure at all, but a defect in the pipe, for which appellant would not be liable under the terms of its policy. We do not consider it necessary, however, to decide that question, in view of our conclusion upon what we have previously stated is the controlling question in the case.

In that case, the court recognized that the mere fact that ammonia gas escapes from an opening in a pipe does not establish that it was the pressure of the refrigerant which caused the opening. Here, we go beyond possible defects in the pipe to positive evidence of corrosion by brine which had reduced the aluminum to tissue thinness in many places, although this received no adequate consideration as a possibility of being the cause of the escape of the ammonia gas. As pointed out in the Prentice case, supra, in such an occurrence, it is incumbent on the plaintiff to exclude by a preponderance of the evidence certain probable causes for which the insurer would not be liable.

Of some significance in this case is the fact that no noise was heard. Noise is constantly referred to in “explosion” cases, and we have been unable to find any in which there had not been some evidence of noise 5 as indicia of violence.

*125 If the formation of the hairline crack, through which the ammonia gas escaped, constituted an “explosion,” no one heard it. It is pointed out that the noise of the operation of the plant and the concrete walls of Room 2 would have made it difficult to hear. However, the expert to whom we have referred testified that in testing another piece of the tubing it took 400 pounds of pressure to develop a hairline crack (exhibit No. 3), and there was only a hissing sound heard simultaneously with the cracking. We are not advised as to how loud the hissing sound may have been, nor as to whether a hairline crack caused by 400 pounds of pressure would make more noise than one caused by 70 pounds or less of pressure. The expert did testify, when asked if there was a loud noise: “No, there wasn’t, I would think; not at all.”

There may have been noise, but it was not heard, so one of the familiar indicia of violence is missing and makes more difficult the determination whether it was the disintegration of the aluminum tubing or the pressure of the ammonia gas which caused the opening through which the gas first escaped.

We are not holding, as a matter of law, that there was no explosion in this case. We are holding that the trial court’s finding of fact, on which the judgment is based, is not sustained by the evidence; and we are holding that the determination of whether or not there was an “explosion” should be made by the trier of the facts on the basis that the term, as used in the insurance policies before the court, is to be construed in its popular sense and as understood by ordinary men and not by scientists.

We are further holding that it should be made clear, by findings or memorandum opinion, that the trier of the facts has considered the likelihood of the opening in the aluminmn tubing having been a continuation of the corroding process which had already reduced the tubing, in places, *126 to tissue thinness. In weighing that theory against the likelihood of an “explosion” caused by pressure within the pipe, it should be made clear what factors remove the cause of the escape of the gas from guess and conjecture!

The basic issue here presented was essentially a question of fact: was there an “explosion” within the purview of the extended coverage endorsements of Minneapolis and New Hampshire. The trial court concluded that there was such an “explosion,” and that may be the ultimate determination; but it is possible that the trial court might have arrived at a different result had it applied the rules which we here hold to be applicable.

The cause is, therefore, remanded to the trial court for reconsideration, in the light of our holdings in this opinion. If the trial court believes that the case should be reopened for further evidence under the circumstances, it may be reopened under such limitations as the trial court may impose.

Costs on this appeal will abide the ultimate determination of the action. : >

Ott, Hunter, and Hale, JJ., and Langenbach, J. Pro Tern., concur.

Each policy provided that: “the coverage of this policy is extended to include direct loss by Windstorm, Hail, Explosion, Riot, Riot Attending a Strike, Civil Commotion, Aircraft, Vehicles, and Smoke.”

There was also the following: “Provisions applicable only to explosion: Loss by explosion shall include direct loss resulting from the explosion of accumulated gases or unconsumed fuel within the firebox (or combustion chamber) of any fired vessel or within the flues or passages which conduct the gases of combustion therefrom. However, this company shall not be liable for loss by explosion, rupture or bursting of:

“(a) Steam boilers, steam pipes, steam turbines or steam engines; or

“(b) Rotating parts of machinery caused by centrifugal force; if owned by, leased by or actually operated under the control of the insured.

“The following are not explosions within the intent or meaning of these provisions:

“(a) Concussion unless caused by explosion,

“(b) Electrical arcing,

“(c) Water Hammer,

“ (d) Rupture or bursting of water pipes.

“Any other explosion clause made a part of this policy is superseded by this endorsement.” ¡

Subsequent to the time this tubing was installed, a different type of aluminum has been developed which is not subject to the same corrosive effect and is used extensively in ships and other places where it is constantly exposed to salt water.

This was the testimony of W. C. Stone, a refrigeration engineer.

As heretofore pointed out, the insurer did limit its liability by specifically designating certain things as not being “explosions” within the intent or meaning of the provisions of the policy.

Couch on Insurance 2d § 42:56, states:

“Noise, although often present in the case of an explosion, is not an essential element of it.”

However, in the lone case which is cited in support of the statement *125 (American Alliance Ins. Co. v. Keleket X-Ray Corp., 248 F.2d 920 (6th Cir. 1957)), a cap was blown from a capsule and there was a popping noise and simultaneous appearance of a cloud of white powder or mist.