Tbe plaintiffs’ appeal presents tbe question of tbe propriety of tbe judgment of involuntary nonsuit rendered by Judge Sharp at tbe close of plaintiffs’ evidence.
In determining this question we are admonished by correct rules of appellate practice to consider tbe evidence in tbe light most favorable for tbe plaintiffs, but observance of this rule does not relieve tbe court of tbe duty of requiring tbat tbe plaintiffs offer some substantial evidence to support tbe allegations of tbe complaint.
The plaintiffs have alleged and staked their action upon tbe allegation tbat their building was struck and damaged by lightning. Conceding tbat no eye saw tbe lightning strike tbe building, tbe plaintiffs nevertheless bring tbe case here for review on tbe theory tbat they have offered circumstantial evidence sufficient to survive a nonsuit and to carry tbe case to tbe jury tbat tbe damage to their building was in fact caused by lightning.
We have examined tbe evidence set out in tbe record, and, without undertaking to quote tbe testimony, we reach tbe conclusion tbat it fails to show more than a possibility or to furnish more than material for conjecture as to tbe cause of tbe damage to plaintiffs’ building. It does not measure up to tbat degree of “reasonable certainty as to probabilities arising from a fair consideration of tbe evidence” tbat would require submission of the issue to the jury. Mfg. Co. v. R. R., 233 N.C. 661, 65 S.E. 2d 379. In the expressive language of Justice Brogden failure to apply this rule “would unloose a jury to wander aimlessly in tbe fields of speculation.” Poovey v. Sugar Co., 191 N.C. 722, 133 S.E. 12; Mfg. Co. v. R. R., 233 N.C. 661 (670), 65 S.E. 2d 379; 32 C.J.S. 1116.
Tbe material facts shown by plaintiffs’ evidence may be briefly summarized as follows: On tbe evening of 16 July, 1951, there was a sudden violent storm, accompanied by lightning and thunder and a downpour of rain, lasting about 20 or 30 minutes. One witness thought there must have been a “cloudburst” of rain, and there were gusts of wind of unusually high velocity. Another witness who bad been at work in a building *760150 feet from plaintiffs’ building, and left in a truck about the time the rain began, testified that before he left he saw a lightning flash and felt the vibration from it and the thunder that followed; that he heard a fire alarm, or bell from a fire truck, from a building across the street; that when he left he drove along the street by the plaintiffs’ building and observed no sign of damage. The next morning about 7:30 a.m. it was discovered that a part of the roof of plaintiffs’ two-story building (100 by 45 feet) had collapsed, and had broken through the second floor and deposited on the first floor a mass of splintered and broken timbers. Between 50 and 75 feet of the roof at the rear, to the width of 45 feet, had fallen in. This part of the roof sloped to the rear. According to a witness it appeared as if some force had struck the roof and “just pushed it through the second floor and on down to'the first floor,” and a lot of water had come in there. The roof was of five-ply felt, with asphalt and gravel, and it was estimated the roof weighed 500 or 600 pounds per 100 square feet. There were some tall trees near the building on the west side, and an electric power line strung on poles paralleled the east side. The building was equipped with electric wiring under the roof, metal flashing and metal downspout.
Plaintiffs noted exception to the ruling of the court in excluding the opinion evidence of the witness Murray "White. This witness, a local insurance agent, had observed the storm from a distance of a quarter of a mile from plaintiffs’ building, and saw the building the following day. He had studied physics at college, had kept up his studies, and as an insurance agent had investigated losses due to lightning. He was tendered by plaintiffs as an “expert in the field of damage by the elements.” The defendants’ objection was sustained. The court did not find he was an expert, and there was no request that the court do so. Stansbury, sec. 133; LaVecchia v. Land Bank, 218 N.C. 35 (41), 9 S.E. 2d 489; Aydlett v. By-Products Co., 215 N.C. 700, 2 S.E. 2d 881; Lumber Co. v. R. R., 151 N.C. 217, 65 S.E. 920. This witness, if permitted to give evidence as an expert would have testified that in his opinion “the damage to this particular building was caused as an indirect result of lightning in a very severe storm . . . And that lightning of a terrible degree of velocity occurred and as a result there was a vacuum created and this vacuum was coming together which caused an unusually high vibration called thunder and because of the impact of that vibration on this roof plus the wind velocity at the time caused the sudden pressure or force of a downward thrust on that roof and caused it to go in at that particular point.” "When asked if he thought the lightning itself directly struck the building he answered “No, I could see no visible evidence where lightning had made contact. It could have been but I could see no evi-*761denee.” It is not perceived tbat tbis witness’ testimony would have been of material help to the plaintiffs. The court’s ruling is upheld.
The evidence in this case leaves the matter in the realm of conjecture and is insufficient to support plaintiffs’ allegation that the building was struck and damaged by lightning.
Plaintiffs call attention to the fact that in the insurance policies sued on the building was insured against direct loss by fire and lightning and that there was extended coverage to include loss from windstorm. But in their complaint the plaintiffs alleged only that the loss resulted from the building being struck by lightning. As was said in Suggs v. Braxton, 227 N.C. 50, 40 S.E. 2d 470, “Plaintiff’s recovery is to be had, if at all, on the theory of the complaint and not otherwise,” and in Moore v. Clark, 235 N.C. 364, 70 S.E. 2d 182, “If the plaintiffs are to succeed at all, they must do so on the case set up in their complaint.” Whichard v. Lipe, 221 N.C. 53, 19 S.E. 2d 14.
We have examined the cases cited by plaintiffs and find they are distinguishable and not controlling on the evidence offered in this case.
The judgment of nonsuit is