The plaintiff had a verdict of $8500 for personal injuries sustained by her as the result of a fall on an icy sidewalk. The defendant has appealed from the judgment thereon, assigning error in the court’s denial of its motion to set aside the verdict and in the charge to the jury.

There was no evidence of actual notice to the defendant city of the alleged defective condition of the walk. It contends that the court erred in refusing to set aside the verdict, because the evidence did not warrant a finding of constructive notice, and because the amount of the verdict was excessive. The jury reasonably could have found the following facts: The plaintiff, who was seventy-five years of age, fell on a public sidewalk on the south side of Hebron 'Street in Hartford at about 11:20 a. m. on November 30, 1948, because of the icy and slippery condition of the walk. There was no sand or other abrasive substance upon the surface. Hebron Street is about two and a quarter miles from the center of *171the city and is in a quite thickly populated residential area. The street runs east and west and has a curb and improved sidewalk on each side. Hartford has a total of 350 miles of sidewalks and employs seven sidewalk inspectors in addition to the police, who are under a duty to observe and report defective sidewalk conditions. There were lumps of ice on the walk, and the surface was icy, very slippery and dangerous. Before her fall the plaintiff was walking carefully between the lumps.

The only testimony as to how long the walk had been in this icy condition, the vital fact upon the issue of constructive notice, was that of the plaintiff’s daughter, Blanche Hedlund. She testified that the ice had been there “in the same condition” for “[a]bout three or four days,” that it “had . . . footprints on it” and was “rough, and thick, and bumpy,” and that the walk for its full width and length “was covered with a sheet of ice.” She also stated she was “sure” that “there had been snow on that sidewalk constantly for three or four days before the accident.” The language quoted suggests the exaggeration of a partisan and interested witness. When this testimony is tested by the undisputed facts shown by the records of the weather bureau which were laid in evidence by the plaintiff, it is manifest that its vice was something more serious than mere overstatement. This is perfectly clear as to the declaration of the witness that she was “sure” that “there had been snow on that sidewalk constantly for three or four days before the accident” happened to the plaintiff on November 30, because the weather bureau records establish unequivocally that the first snowfall of the season occurred on November 29. The bureau’s recordings of temperature and precipitation for the period *172prior to November 30, which are recited below, indicate with almost equal certainty that the witness’ statements that ice such as she described covered the walk during these three or four days before the plaintiff’s fall were also contrary to fact. Whether acceptance by the jury of this testimony as credible evidence would be so unreasonable, and so suggestive of a verdict produced by improper influence, as to call for setting it aside under the principle of Roma v. Thames River Specialties Co., 90 Conn. 18, 20, 96 A. 169, it is unnecessary to determine, for reasons which we proceed to consider.

It is a sound and well-established principle of our law that “[w]hen testimony is ... in conflict with indisputable physical facts, the facts demonstrate that the testimony is either intentionally or unintentionally untrue, and leave no real question of conflict of evidence for the jury concerning which reasonable minds could reasonably differ.” Mlynar v. A. H. Merriman & Sons, Inc., 114 Conn. 647, 650, 159 A. 658; Richard v. New York, N. H. & H. R. Co., 104 Conn. 229, 232, 132 A. 451; Jump v. Ensign-Bickford Co., 117 Conn. 110, 115, 167 A. 90; Horvath v. Tontini, 126 Conn. 462, 464, 11 A. 2d 846. We apply this principle to determine whether the indisputable physical facts established by the records of the weather bureau demonstrate that the testimony of Blanche Hedlund was untrue and so, as a matter of law, could not constitute sufficient evidence to support a finding by the jury that the city had the constructive notice essential to sustain the plaintiff’s verdict. The city could not be held liable unless the condition had existed for such a length of time that it was charged with notice thereof and had had a reasonable time thereafter to remedy the defect. Ritter v. Shelton, 105 Conn. 447, 450, 135 A. 535.

*173The weather bureau records show that the temperatures for the month of November, 1948, averaged substantially above normal and that the rainfall was unusually heavy, totaling nearly seven inches. The grades and the terrain were not such as to cause rainfall to collect and remain on the walk. In determining the probative effect of the weather bureau’s recordings with relation to the formation of ice upon the walk, it is the essential coincidence of precipitation and freezing temperatures which is decisive. The daily record of temperatures and precipitation leaves no possible doubt that as of November 24, 1948, there was neither ice nor snow on the sidewalk where the plaintiff’s fall occurred on November 30. For the seven consecutive days ending with November 30, the daily temperature ranges in degrees Fahrenheit were: Nov. 24, 31-50; Nov. 25, 32-52; Nov. 26, 25-52; Nov. 27, 40-59; Nov. 28, 36-48; Nov. 29, 24-40; Nov. 30, 22-38. The only precipitation during this period was .29 of an inch on the 27th and .48 of an inch on the 29th, on which day the first snowfall of the season occurred. Inasmuch as there was no precipitation between the 23d and the 27th, when there was .29 of an inch, no ice could have formed before the 27th. Since at no time on the 27th and 28th did the temperature drop below 36 degrees, no ice could have formed before 7 a. m. on the 29th, when the temperature touched 32 degrees for the first time. That morning the precipitation between 3 and 5 o’clock was .07 of an inch in the form of rain; between 5 and 6 o’clock, .08 of an inch in the form of mixed snow and rain; between 6 o’clock and 12 noon, .32 of an inch in the form of snow; and in the first hour of the afternoon, between 12 and 1 o’clock, .01 of an inch in the form of snow. This made a total of .48 of an *174inch on the 29th. There was no further measurable precipitation on that day, and there was none on the 30th. It therefore is clear that the indisputable physical facts conclusively determine that the testimony of Blanche Hedlund did not constitute credible evidence upon which the defendant could be charged with constructive notice.

Although there was no testimony by any other witness as to What the condition of the walk was before the plaintiff fell, the question remains whether the weather bureau records, considered in connection with the testimony as to the icy condition existing when she fell, constitute circumstantial evidence sufficient to warrant a finding by the jury of constructive notice. See Gleba v. New Britain, 133 Conn. 85, 87, 48 A.2d 227. The facts shown by these records determine that the maximum possible period during which there could have been ice on the walk was from 7 a. m. on the 29th to 11:20 a. m. on the 30th, when her fall occurred, a total of about twenty-nine hours, and that during this interval the température did not rise above 32 degrees, except that between 12 noon and 4 p. m. it touched a maximum of 35 degrees. This does not mean, however, that there was evidence warranting a finding by the jury that the slippery condition which caused the plaintiff’s fall had existed for either twenty-nine hours or any substantial part thereof. In fact the positive evidence very strongly indicates the contrary.

Police officer Laban visited the locus shortly after the accident. His uncontradicted testimony was: “The sidewalk had been shoveled. There was snow on each side of the sidewalk where it was thrown from the shovel, and the ice formed on the sidewalk appeared to have come from fallen snow from the *175trees.” The testimony of the meteorologist, based on the weather bureau records, was that at 7:30 p. m. on the 29th “there was one inch of snow on the ground.” Aside from the testimony of the plaintiff’s daughter, which for the reasons already stated is not to be considered, there is not a syllable of testimony that there was any snow upon the walk at the time the plaintiff fell. On the contrary, the plaintiff herself and three other witnesses, the only other ones to testify to the walk’s condition, stated in substance that the surface consisted of very slippery glare ice with some bumpy places in it. Since, after 7 a. m. on the 29th, the temperature had continuously remained near freezing or below, the one inch of snow which had accumulated on the walk by 1 o’clock that afternoon could not have melted to form the ice upon which the plaintiff slipped. In other words, the jury could only have concluded that the slippery condition which caused her fall was created by the removal of the snow from the walk, which thus exposed the surface beneath whereon the precipitation of rain and of mixed rain and snow had lodged between 3 and 6 a. m. on the 29th.

There is no evidence to show, however, when the snow was shoveled from the walk. From aught that appears, this may have been late on the night of the 29th or even early on the morning of the 30th. On this state of the record there was no evidence upon which the jury could reasonably have found how long the defective condition that caused the plaintiff’s fall had existed. The burden rested upon her to prove this fact, for it was essential in order to charge the defendant with constructive notice. In determining whether a ease has been made out in an action of this nature, certain guiding principles are to be kept in mind. What this court stated many *176years ago still holds true: “[I]n this rigorous climate the duty of cities and towns in respect to snow and ice is and must be very limited.” Congdon v. Norwich, 37 Conn. 414, 419; Cloughessey v. Waterbury, 51 Conn. 405, 418; Ross v. Stamford, 88 Conn. 260, 261, 91 A. 201; Carl v. New Haven, 93 Conn. 622, 625, 107 A. 502. “[I]n our climate, considerable latitude should be allowed municipalities in cases of this character . . . .” DeCrosta v. New Haven, 119 Conn. 344, 347, 176 A. 268. Even where the highway defect consisted of a break in a sidewalk and was not one arising from snow or ice, we held in a case tried to the court that a finding that the defect “had existed for a considerable time, but just how long is not known” was “too indefinite to permit us to determine whether the break had existed long enough to have charged the defendant with its knowledge or not.” Burgess v. Plainville, 101 Conn. 68, 72, 124 A. 829. In the instant case, a finding by the jury that the slippery condition which caused the plaintiff’s fall “had existed for a considerable time” would exhaust the most favorable inference which they could reasonably draw from the evidence. Their finding that the defendant city was chargeable with constructive notice was therefore unwarranted and the verdict cannot stand. See Frohlich v. New Haven, 116 Conn. 74, 75, 163 A. 463; Bazinet v. Hartford, 135 Conn. 484, 488, 66 A.2d 117; Jainchill v. Schwartz, 116 Conn. 522, 524, 165 A. 689.

'The weather bureau is located at Brainard Field in Hartford, about four miles from Hebron Street. The possible slight variations in temperature and precipitation at the two locations are not such as can materially affect our conclusion. See Lessow v. Sherry, 133 Conn. 350, 354, 51 A.2d 49. While we refused in that case to disturb the plaintiff’s ver*177diet in response to the defendant’s claim that the weather burean records furnished indisputable physical facts which required doing so, the difference in the factual situation clearly distinguishes the case from the one now before us. The same holds true as to the case of Tenney v. Pleasant Realty Corporation, 136 Conn. 325, 328, 70 A.2d 138. The principal injury sustained by the plaintiff was a broken wrist. In view of our conclusion that the court erred in denying the motion to set aside the $8500 verdict in her favor, it is unnecessary to consider the defendant’s further claim that the verdict was excessive. The same holds true as to the error assigned in the court’s charge to the jury.

There is error, the judgment is set aside and a new trial is ordered.

In this opinion Jennings and Baldwin, Js., concurred.

(dissenting). Just one week ago we released an opinion wherein we reiterated a principle to which we have always adhered. “Nothing in our law is more elementary,” we said, “than that the trier is the final judge of the credibility of witnesses and of the weight to be accorded their testimony.” Morgan v. Hill, 139 Conn. 159, 161, 90 A.2d 641. The majority opinion, it seems to me, undermines this principle.

My associates conclude that the indisputable facts, disclosed by the weather reports, establish the inaccuracy of any testimony that the walk in question had been covered with ice for three or four days. I am in accord with that conclusion. I disagree, however, with the assertion that there was no evidence of the length of time during which the walk was icy. When a witness testifies that a defective *178condition bad existed for a specified number of days, the jury are not bound to accept or reject the testimony in toto. If circumstances warrant, they may conclude that the witness is mistaken only as to the length of time stated. If deemed proper, they may use his testimony to find that the condition had existed for a lesser time. .

As applied to this case, these principles would permit the jury to accept the testimony of the plaintiff’s daughter that the walk was icy and to discount to one day the length of time it had been in that condition. With such evidence, we cannot hold, as a matter of law, that the city did not have constructive notice of the icy walk for a sufficiently long time before the accident to have remedied it through the use of reasonable care.

In this opinion Inglis, J., concurred.