Frank T. Gallagher, Justice.

Certiorari to review an order of the industrial commission denying compensation.

*272 Eelator is the widow of Edward T. Sullivan, who died on January 16, 1952, at St. Joseph’s Hospital in Brainerd at the age of 44. Sullivan was hired by Hagstrom Construction Company and commenced work on December 4,1951, working intermittently as a night watchman, although his primary duties consisted of tending the salamander stoves used by employer on its construction project. On December 12, 1951, he went to work at 4 p. m. and worked a double shift ending at 7 a. m. the following morning. He then drove his car approximately three miles to his home, arriving there at about 9:30 a. m. Petitioner testified that he was sick to his stomach when he arrived home that morning and went directly to bed without eating; that he vomited several times during the day; that his face and hands were very warm; and that he could not stand up without assistance. That evening employee was taken to the hospital, where a Dr. Cardie gave him medication and advised him to return home and go to bed. Employee did so but gained no relief from the medication. He was restless and could not sleep, and he vomited and tossed in his bed all night.

The next afternoon employee was taken to Dr. William E. Fitzsimons, who examined him and prescribed some medication. Employee took the medicine, but his vomiting continued. The following day, December 15, employee was admitted to the hospital, where he gave the following history to Dr. Fitzsimons: He had been tending salamander oil burners on employer’s construction project; .during the night of December 12-13 one of the salamanders went dry and exploded; the gas he was breathing was extremely blue; and he had a hard time getting away from the smoke, having almost lost consciousness. In referring to his hospital notes while testifying, Dr. Fitzsimons said that employee told him that the smoke choked him and made it extremely hard for him to get his breath and that he had difficulty in getting out of the area in which he was located while he was near this oil burner.

On employee’s admission to the hospital, Dr. Fitzsimons noted for the hospital record that employee complained of severe headaches and dizziness and that he had extreme pallor and nausea in addition *273 to elevated blood pressure, extreme apprehension, and vomiting. His vomiting continued when he was in the hospital, and Dr. Fitzsimons ordered nasal suction to empty the contents of employee’s stomach. Suction was in place on December 21 and 22, and employee’s condition improved and he felt much better. He was allowed to go home on December 24 for the Christmas holidays, but petitioner testified that he was in bed from the evening of December 24 through 27, when he was readmitted to the hospital. During his brief stay at home, employee began vomiting again, and, when readmitted, he complained of nausea, severe headache, dizziness, and also visual disturbances. At this time Dr. Fitzsimons diagnosed his condition as acute glomerulonephritis with uremia. The effect of this condition was described by Dr. Fitzsimons as a blocking of the passage of blood through the kidneys, thereby causing increased blood pressure and inability of the kidneys to filter the waste products from the blood and pass them out through the urinary system. From December 27 employee’s condition grew progressively worse until his death on January 16, 1952.

The doctors who testified at the hearing agreed that the immediate cause of death was uremia and that the basic underlying cause of uremia was glomerulonephritis, but they were disagreed as to whether that disease was acute or chronic. The evidence from the autopsy performed on employee and the testimony of the doctors indicated that employee was suffering from nephritis prior to his employment by employer.

The question raised upon appeal is whether employee sustained an accidental injury to his person on or about December 13,1951, which contributed to his death on January 16, 1952, and, if so, did the injury arise out of and in the course of his employment.

The salamander-type oil burners tended by employee on the night of December 12-13 were on the third-floor level of a building under construction in a room about 75 feet by 75 feet enclosed around the sides by canvas. The room was enclosed on top, 12 feet above the third-floor level, by a slab of newly poured concrete. The salamanders were producing heat to maintain a temperature of about 75 to 85 degrees in order to allow the newly poured concrete to set and cure *274 properly despite the winter weather. The salamanders were about five feet high, including the smokestack, and each one contained about ten gallons of No. 1 fuel oil. Employee’s duties were to refuel and keep burning the 16 to 18 salamander stoves within the enclosed area. The salamanders gave off a blue smoke as they burned the fuel oil, and they were not vented to the outside air. On the contrary, the smoke was released within the enclosed area. There is testimony that the room was very smoky and that no one could work inside the canvased area for more than an hour or so. It appears that the only place the smoke could escape was through the rips or breaks and tears in the canvas and the narrow cracks around the top. There is also testimony from some of the workmen in the area that their eyes smarted, that their throats would get dry, and that they would cough while in the enclosure. The men who tended these salamanders were supplied by employer with flashlights to help them find their way around inside the enclosed area, as it appears from some of the testimony that the workmen could see each other only from a distance of from six to ten feet inasmuch as the electric lights were clouded by the smoke.

The referee found that on the date involved employee sustained an accidental injury arising out of and in the course of his employment; that as a result of said accidental injury employee became temporarily totally disabled on December 13, 1951, and continued to be so disabled until January 16, 1952, the date of his death; that said accidental injury contributed to his death; and that at the time of his death employee left surviving him a widow and minor child, who were dependents within the provisions of the workmen’s compensation law of this state.

The commission did not agree with the referee and held on appeal that the determination of the referee was not in accord with the evidence and the law and thereupon vacated and set aside the decision of the referee. It was the opinion of the commission that employee at no time relevant herein suffered any accidental injury to his person which resulted in his death on January 16, 1952, and that employee’s disability subsequent to December 13,1951, and his death on *275 January 16,1952, were not the result of an accidental injury arising out of and in the course of his employment.

We recognize that the question before us is a close one as evidenced by the differences in the opinions of the referee and the commission. The commission decided that there was no evidence that there was any carbon monoxide present; that there was no factual foundation for the assumption that monoxide poisoning caused or aggravated the condition which resulted in employee’s death; and that the medical evidence in support of petitioner’s claim was wholly speculative as to causal relationship.

We cannot agree that there was no evidence of any carbon monoxide present. While it is true that there was some textbook reference in the briefs to the effect that carbon monoxide is a colorless and odorless gas, there was also some showing to the effect that such fumes and heavy blue smoke as existed in the enclosure on the night in question contained carbon monoxide. For example, Dr. George Berdez, the pathologist who performed the autopsy on the body of employee, was asked:

“Q. * * * what part of the fumes is it that might have caused some deleterious effect upon Mr. Sullivan, is it carbon monoxide?

“A. Such fumes contain carbon monoxide and carbon monoxide might have a deleterious effect, and there might be other substances in those fumes, but I would not know which.”

At another point Dr. Fitzsimons testified:

“* * * The history of this man is that he was overcome, at least partially, by smoke, carbonmonoxide, of course, and from the combustion of either a dry or exploding tank, I don’t know.”

When questioned again on cross-examination, Dr. Fitzsimons said:

“Q. You don’t know what, if anything, it [salamander] generated by way of gas?

“A. I know that fuel oil combustion forms carbonmonoxide.”

The commission apparently put considerable emphasis on Dr. Fitzsimons’ statement in his testimony that his patient definitely died of uremic poisoning caused by nephritis and that he did not know what *276 effect the fuel oil combustion had on it. While it is true that Dr. Fitzsimons made such a statement, he also said that he felt that the uremic condition had existed somewhat previously and that he could not say whether the lack of oxygen caused by the explosion affected that condition except that the patient developed a more acute illness after the accident. The doctor then said that “it seems logical that some effect, at least, aggravated the already previously existing condition and caused a premature termination of his condition or his nephritis.”

In a letter dated September 26, 1952, to D. B. Bumble, attorney for employer, which was made a part of the record, Dr. John B. Moyer, who had examined employee at the hospital in Brainerd the day before his death, concluded that some concentration of carbon monoxide very likely was present in the enclosure, possibly in sufficient quantities to produce symptoms over a sufficient period of time. It is interesting to note that Dr. Moyer, at the time of the examination of employee, diagnosed his condition as acute nephritis, and, while his report of the examination noted that the relationship between the blast injury of December IB, 1951, and his condition at the time of the examination was not clear, the report stated that the evidence strongly pointed to some relationship because of lack of symptoms existing prior to the injury. However, Dr. Moyer testified, after reading the autopsy report of Dr. Berdez, that in his opinion carbon monoxide had nothing to do with employee’s death.

Dr. J. P. Korchik, who was called as an expert witness by petitioner, was asked:

“Q. Does the evidence here of fumes being present indicate necessarily that carbon monoxide is present?

“A. As I have stated frequently, I don’t know how much carbon monoxide was present; I have to assume that it was a considerable amount of carbon monoxide.”

We cannot understand, in the light of the above testimony, how the commission could say that there was no evidence at all under the record here that any carbon monoxide was present. It seems to us that, viewing the record in its entirety, it would be difficult to say *277 that there was no evidence at all of any carbon monoxide being present. We cannot agree with the commission that there was no factual foundation for the assumption that monoxide poisoning caused or aggravated the condition which resulted in employee’s death nor that the medical evidence in support of petitioner’s claim was wholly speculative as to causal relationship.

Without attempting to review in detail all the medical testimony in this respect, it is very evident from an examination of the record that the commission was most impressed with the positive testimony of Dr. John F. Briggs and Dr. Charles N. Hensel as to the lack of causal relationship. Those two doctors, neither of whom ever saw employee, were called by employer and testified in answer to hypothetical questions that there was no causal relationship between employee’s death and his working conditions. Dr. Hensel stated with reference to the presence of carbon monoxide in the canvased area that “there is no evidence in the record that there was any explosion that night, so you only have his [employee’s statement to Dr. Fitzsimons] hearsay testimony for that.” Dr. Hensel admitted that he had never seen a salamander stove although he had heard the testimony of employer’s foreman describing a salamander and its method of combustion.

While we are cognizant of the well-established rule that this court cannot resolve conflicts in medical testimony, we cannot ignore the testimony of both Dr. Fitzsimons and Dr. Korchik as to causal relationship, as it seems to us that there was an adequate factual foundation for their testimony. In addition to the testimony of Dr. Fitzsimons referred to above, Dr. Korchik said, when asked by the referee for his unadulterated opinion, uninfluenced by the opinion of any other physicians who examined employee either before or after death but taking into consideration all information contained in the autopsy report:

“* * * This patient was suffering from glomerulonephritis and from uremia, hypertension and all the pathological effects which result thereof, but in addition to that I feel that the effects of fuel oil combustion poisons which contained carbon monoxide aggravated *278 Ms illness and to a certain degree was responsible for Ms death. * * * in my opinion this man died from the effects of glomerulonephritis, uremia, and from the effects of carbon monoxide poisoning.” (Italics supplied.)

In Saaf v. Duluth Police Pension Relief Assn. 240 Minn. 60, 59 N. W. (2d) 883, this court said that, in order to have a proper foundation for a finding of causal connection in cases where such connection must be established solely by expert testimony, the medical expert must, upon an adequate factual foundation, testify not only that in his professional opinion the injury in question might have caused or contributed to the subsequent death of the injured person but further that such injury did cause or contribute to his death. However, the court said in that case that such medical testimony need not be expressed in any particular words. All that is needed, as applied to the facts of a particular case, is that it appears from the testimony of the medical witness as a whole that he is of the opinion that the injury not only could have caused the person’s death but did cause it. It was further said in that case that the opinion of the expert did not have to express an absolute certainty in the matter.

Again in Hiber v. City of St. Paul, 219 Minn. 87, 16 N. W. (2d) 878, we said with reference to an expert’s opinion that it is not necessary that the truth of such an opinion be capable of demonstration but that it is sufficient if it probably is true. He is not required to speak with such confidence as to exclude all doubts in his mind but may render his testimony in the form of an estimate of opinion couched in expressions which may fall short of absolute conviction of accuracy.

It seems to us that, looking at the testimony of Doctors Fitzsimons and Korchik as a whole, they expressed the opinion that there was a presence of carbon monoxide in the case at bar and also that it not only could have but did aggravate a condition which hastened employee’s death. For the commission to have more or less disregarded such testimony appears to us to have been a misapplication of the rules stated above. For these reasons, we believe that the case should *279 be remanded for a rehearing and for the introduction of such additional evidence as the parties may desire to introduce and for a reconsideration on the part of the commission in the light of this opinion.

Relator is allowed $250 attorneys’ fees and costs and disbursements herein.

Reversed and remanded.