delivered the opinion of the court:
This is a direct appeal brought by complainant, Royal Elm Nursing and Convalescent Center, Inc., from an order entered by the Illinois Commerce Commission (the Commission), denying Royal Elm’s complaint against respondent, Northern Illinois Gas Company (NI-G), which arose out of a bill issued to Royal Elm by NI-G for gas allegedly used during the period of June 18, 1976, through March 14, 1983, but which was unmetered due to a tampered gas meter. In its complaint, Royal Elm denied that it owed NI-G payment for unme*76tered gas and claimed that NI-G’s claim had not been substantiated and was not well-founded. The sole issue on appeal is whether the findings of the Commission as to the date the meter tampering occurred are supported by substantial evidence. For the following reasons, we reverse the Commission’s order and remand the cause with directions.
The record sets forth the following facts relevant to this appeal. Royal Elm, located at 7733 Grand Avenue, Elmwood Park, Illinois, commenced operation of its nursing home and convalescent facility in 1975. Rudolph Tessler, Royal Elm’s president, testified that the facility’s occupancy rate fluctuated over the years until in 1983 it was approximately 80% occupied. In March 1983, as part of a routine service check, NI-G removed the gas meter index from the outside of Royal Elm’s facility and replaced it with a new tamper-resistant model. It was NI-G’s policy to remove gas meter indexes every 10 years for testing. When the old meter was tested, the results indicated that the meter index had not been registering all of the gas passing through it. In fact, 41/2 of the 16 teeth of a critical gear wheel had been cut off, causing the meter index to register only about 75% of the actual gas usage. Because there was no visible means to detect tampering, NI-G had had no indication that anything was wrong with the meter until it had been tested.
Pursuant to Commission Rule 280.100(c) (83 Ill. Adm. Code 280.100(c) (1985)), once it has been discovered that “there has been tampering with wires, pipes, meters or other service equipment and the customer has enjoyed the benefit of such tampering, *** [t]he customer shall be responsible for all service usage *** during the period tampering occurred.” Thus, once tampering is discovered, it is irrelevant who did the tampering. The responsibility for rectifying the underpayment is automatically placed on the gas consumer. In the present ease, it is undisputed that someone had tampered with the gas meter index which had been removed from the outside of Royal Elm’s facility in March 1983. However, Royal Elm disputes the Commission’s finding as to when the tampering occurred on the grounds that it is not supported by substantial evidence in the record.
Lynwood Valor, director of measurement for NI-G, testified at the hearing that, by all outward appearances, the tampered meter index moved and functioned as if everything were operating satisfactorily. The error occurred when the measurement of gas usage was transmitted to the index. The measurement was reduced in direct proportion to the number of teeth removed from that particular gear. In Royal Elm’s case, 41/2 out of 16 teeth were missing. Thus, there was *77approximately a 25% reduction in actual measurement.
Norbert Oliver, director of marketing for NI-G, testified as to the analytical method used to determine when the tampering had occurred. Initially, Oliver compiled all of the billing data available for Royal Elm and analyzed this data to obtain an estimate of: (1) use per degree day during the heating season and (2) base use per day during the nonheating season. Oliver then prepared an account analysis which compared the amount of gas usage measured at Royal Elm for the summer periods and winter periods from June 1975 through June 1984. Specifically, Oliver looked for an indicator of a sudden decrease in both use per degree day (winter) and base use (summer) roughly equal to the meter reading error of 25%. Oliver located a decrease in both uses between the 1975 and 1976 readings when the use per degree day dropped 16.6% and the base use dropped 27.6%. Based on his analysis, Oliver estimated that the tampering had occurred close to the end of the 1975-76 heating season. Thus, the reading date taken at the end of that heating season on June 18, 1976, was determined to be the date at which the unmetered gas usage commenced. In order to determine the amount owed for unmetered gas, Oliver adjusted the billing previously issued to Royal Elm for the period from June 18, 1976, through March 14, 1983, the date of meter exchange, by a factor derived from the 25% reduction in gas usage, resulting in a bill to Royal Elm of $45,994.35 for unmetered gas.
Royal Elm filed its complaint against NI-G on May 16, 1984. A hearing was held before the Commission on September 20, 1984, and a final order was entered on October 29, 1986. The findings of fact made by the Commission were as follows:
(1) Royal Elm’s gas meter index was damaged by a deliberate act of tampering at or shortly before June 18, 1976, and Royal Elm was responsible for that tampering;
(2) Best estimate of unmetered gas usage: 160,196.53 therms;
(3) Royal Elm was responsible for the tampering, thus held responsible for the costs;
(4) NI-G’s bill of $45,994.35 was correct;
(5) Royal Elm’s complaint denied.
Royal Elm’s appeal followed.
Initially, Royal Elm contends that because NI-G’s claim alleged the criminal offense of tampering, NI-G has the burden of proving its claim by clear and convincing evidence. We disagree for the fundamental reason that NI-G never filed an action against Royal Elm alleging that Royal Elm had tampered with the meter index. In fact, *78NI-G never filed any action against Royal Elm. Instead, based on its discovery that Royal Elm’s meter index had been tampered with, NIG simply billed Royal Elm for the unmetered gas usage. As indicated, pursuant to Commission Rule 280.100(c) (1985), NI-G had to establish only that there was tampering and that the customer benefitted from that tampering in order to bill the customer for all service usage during the period that the tampered meter remained in use. The question of guilt or innocence never arises.
With respect to this court’s scope of review of an order issued by the Commission, sections 10 — 201(d) and (e) of the Illinois Public Utilities Act (Ill. Rev. Stat. 1985, ch. lll2/s, pars. 10 — 201(d), (e)) state, in pertinent part:
“(d) *** The findings and conclusions of the Commission on questions of fact shall be held prima facie to be true and as found by the Commission; rules, regulations, orders or decisions of the Commission shall be held to be prima facie reasonable, and the burden of proof upon all issues raised by the appeal shall be upon the person or corporation appealing from such rules, regulations, orders or decisions.
(e) Powers and duties of Reviewing Court:
* * *
(iv) The court shall reverse a Commission rule, regulation, order or decision, in whole or in part, if it finds that:
(A) The findings of the Commission are not supported by substantial evidence based on the entire record of evidence presented to or before the Commission for and against such rule, regulation, order or decision ***.”
In determining whether evidence is substantial, the reviewing court must view the evidence in connection with all the circumstances of the case which either add or subtract from the credibility of the evidence. (Wilkey v. Illinois Racing Board (1978), 65 Ill. App. 3d 534, 381 N.E.2d 1380.) In the present case, the only evidence introduced to determine when the tampering occurred was the account analysis prepared and interpreted by NI-G’s expert witness, Norbert Oliver, director of measurement for NI-G. When Oliver’s testimony regarding the account analysis is viewed in connection with all the circumstances of the case as elicited by the testimony at the hearing, it is our opinion that the credibility of Oliver’s interpretation of the account analysis and his conclusion as to when the tampering occurred are severely undermined for the following reasons: (1) determination of the tampering date was predicated on (a) circumstantial evidence from which inconsistent inferences can be drawn, and (b) educated *79conjecture based on the implied assumption that all other factors remained constant; and (2) the account analysis is incapable of being rebutted.
Although circumstantial evidence will generally suffice whenever an inference may reasonably be drawn from it (Mort v. Walker (1983), 98 Ill. 2d 391, 457 N.E.2d 18), the existence of a fact may not be inferred from the evidence when the existence of another fact inconsistent with the first can be inferred with equal certainty from the same evidence. Further, mere conjecture, guess or suspicion is insufficient to support a conclusion drawn from circumstantial evidence. (Lewis v. Mount Greenwood Bank (1980), 91 Ill. App. 3d 481, 414 N.E.2d 1079.) This is particularly true with respect to circumstantial evidence testified to by an expert witness. Regardless of how skilled or experienced an expert may be, he is not permitted to speculate or to state a judgment based on conjecture, i.e., a conclusion based on assumptions not in evidence or contradicted by the evidence. Boose v. Digate (1969), 107 Ill. App. 2d 418, 246 N.E.2d 50.
In the present case, although Oliver recognized that other factors such as patient load, heating patterns, and laundry practices may have precipitated the drops in Royal Elm’s gas usage, he completely ignored these factors in reaching his determination. Thus, in effect, Oliver based his conclusion on the unsupported assumption that all other factors were constant. In its order, the Commission stated that it was “unreasonable to require [NI-G] to obtain such information from its customer before billing for unmetered service, where a reasonable estimate can be made from a billing history and other available information.” In our view, it is even more unreasonable to recognize the likely existence of other factors, yet ignore them, and shift the burden to the consumer to prove that NI-G’s data is erroneous. NI-G’s selective data compilation is apparent in the following colloquy which ensued following a question to Oliver as to whether factors other than tampering could have accounted for the drop in Royal Elm’s gas usage:
“OLIVER: If we look at thermostat settings, changes in equipment, changes in occupancy, changes in insulation, changes in any number of factors, any one of them, all other things being equal, would be expected to have an effect on the usage per degree day of the customer. If it’s something which affects usage during the heating season, if it’s a constant usage on a year-round basis, such as the addition of a dryer or the deletion of a gas dryer, that could be expected to have an effect on the base use per degree day of the customer, all other *80things being equal.
COUNSEL: Those factors which you have just enumerated, however, you would have no knowledge of?
OLIVER: I have no knowledge of them. The only things I have knowledge of are the changes in the customer’s pattern of consumption and usage either in the summer period or in the heating period.
COUNSEL: So you, of course, could not consider in your analysis such factors as occupancy of the building, such factors as what appliances might have been in use consistently during one period and not used during another; such as thermostat settings made by the customer; is that correct?
OLIVER: My analysis does not include any reflection of any of those individual items, except that I might point out that the increasing use per degree day from the 1976 level seems logical related to [the earlier] description of [Royal Elm’s] occupancy level.
COUNSEL: But the factors were not considered in your analysis which resulted in the creation of [the account analysis], were they?
OLIVER: No, sir.”
Later, when queried as to the causes of other isolated drops in either use per degree day or base use, Oliver answered, “I don’t know what caused anything,” and reiterated that his conclusion as to the tampering date was based solely on the simultaneous drop in both indicators. Subsequently, after Oliver answered a question posed by the examiner as to the difference in gas usage between water heaters and clothes dryers, counsel for Royal Elm asked Oliver how he had arrived at the mathematical estimate given to the examiner. Oliver replied that it was “a subjective estimate, very subjective” He then admitted that he had no “factual information” as to the question asked, but had merely given a “subjective answer.”
In our view, Oliver’s testimony is replete with factual conclusions inferred from evidence that readily leads to inconsistent factual conclusions and is impliedly grounded on the contradicted assumption that, at the time both indicators dropped, “all other things were equal.” This assumption is not supported by the evidence. In fact, the evidence clearly indicates that there were a multiple of variables that could have altered both uses. Further, we are not persuaded that the simultaneous drop of 16.6% in use per degree day and the 27.6% drop in base use is substantial evidence of a tampering defect which resulted in an underregistration of 25%. In our view, it appears that *81once tampering was discovered, NI-G settled for any simultaneous drop in indicators it could find as evidence of the date of the tampering, without any investigation as to the impact of other factors. While 27.6% has a reasonable correlation to 25%, we do not find that 16.6% has that same degree of reasonableness. In our view, this discrepancy further adds to the speculative nature of NI-G’s evidence.
In arguing for the validity of its conclusion as to when the tampering had occurred, NI-G contends that because Royal Elm had failed to produce evidence which was solely within its control as to other factors which may have caused the simultaneous decrease in indicators, the presumption arises that the withheld evidence would be unfavorable to Royal Elm. We agree with NI-G that this presumption arises when available evidence is not produced by a defendant (Fuery v. Rego Co. (1979), 71 Ill. App. 3d 739, 390 N.E.2d 97), but we do not find that presumption applicable to the circumstances at bar. In order for the presumption to arise, the evidence must have been available. In the present case, there is no evidence that records as to ventilation, insulation, cooking loads, hot water usage, temperature settings, settings of air dampers and other possible variables were available at the time of the hearing or available at all. Further, Royal Elm was under no obligation to maintain such records and had no notice it should do so. Although Morris Gamze, Royal Elm’s expert, stated that there had not been time to “agglomerate that kind of data,” we do not find that that statement indicates data was available and not brought forth. The only evidence that was conclusively available pertained to census reports, which Royal Elm’s president testified to by giving approximate occupancy figures.
Finally, with respect to the account analysis prepared by NI-G from data accumulated by NI-G and interpreted by an NI-G employee, we note the inability of a layperson or his own expert to verify the accuracy of the data on the analysis. Even in the unlikely event an NI-G customer kept copies of all his gas bills for the 10-year period a gas meter index is in place, it would be difficult, if not impossible, to correlate the figures on those bills with the NI-G data. In essence, Commission Rule 280.100(c) imposes “responsibility” for payment of underregistered gas usage on the person or entity who has benefitted from the underregistration without any means for the customer to dispute figures used in arriving at his usage or in verifying any increases or decreases. Considering that underregistration could occur for as long as 10 years before it is discovered by a meter index exchange, this “responsibility” could result in an onerous financial burden to an unsuspecting user. The possible inequities of this rule are *82readily apparent. For example, a unit owner in a condominium which has individual gas meters could tamper with all the meters in the building so as not to draw attention to his own. During the tampered period, units are bought and sold with no knowledge that underregistration is occurring. The question arises as to who bears the “responsibility” as to what could be a sizeable financial burden when the tampering is eventually discovered. In our view, if the utility is going to place this burden on the consumer, it should, in the very least, place tamper-resistant gas meters inside buildings where they are not readily accessible to nonowners and should install a new gas meter index every time ownership of the building or condominium unit changes.
While we recognize that utility companies are not in the business of providing free service, we are also cognizant of the age-old legal maxim that liability must be established before a penalty is imposed. NI-G correctly argues that it has not alleged criminal conduct. Pursuant to Rule 280.100(c), that allegation is unnecessary because liability or guilt is subsumed under the guise of “responsibility.” We further note that once a consumer is found “responsible” for the unmetered gas and files a complaint with the Commission or with the court, he then assumes the legal status of complainant or plaintiff and has the burden of proving his own innocence. Bearing this in mind, although the issue of the constitutionality of Commission Rule 280.100(c) is not before this court, we question the Rule’s impact on the due process rights of NI-G customers.
For the aforementioned reasons, we conclude that the findings of the Commission are not supported by substantial evidence, and, accordingly, we reverse the Commission’s order, and remand the cause to the Commission with directions to enter a finding in favor of Royal Elm.
Reversed and remanded.
O’CONNOR and MANNING, JJ., concur.