Fairness like other social norms is usually stabilized by punishing norm deviations. Reward uncertainty, however, questions that norm deviations are detected and thus punished. By investing in information acquisition a responder in an ultimatum experiment determines endogenously whether unfair offers are detected and sanctionable. In our experiment a proposer and a responder can share 12 black and 12 white chips where the monetary value of a white chip for the proposer can be rather high ('high payoff mode') or low ('low payoff mode'). The responder can buy information about the proposer's reward type what results in commonly known monetary rewards. The efficient (Pareto-optimal) allocation of chips needs one-sided allocation of - at least one kind of - chips. At the same time Pareto-efficiency is fully compatible with equity. The efficient equitable allocation in the case of high (low) payoff mode gives all black chips to the responder (proposer) and splits white chips equally. We expected lack of information to yield much of waste and inequitable allocations. This expectation is based on the proposer who in the lack of common knowledge does not do the rational differentiation in order to avoid misunderstandings, but relies more on primitive fairness norms (splits both piles of chips equally). According to our results 1. More than half of the responders do not buy information about the payoff mode (30 out of 55). 2. Those, who buy information report more uncertainty about the allocation problem and trust less their proposers. 3. Proposers, whose partners asked for information: - utilize the resource of white chips to a less extent in both payoff modes; - offer considerably less Pareto-optimal allocation in both payoff modes; - offer more equitable allocation in the high payoff mode than those with partners not asking for information. In general, in our experiment buying information did not pay on average (although it was sold at a low price, equal to the value of a single chip). Surprisingly, buying information destroyed rationality in the offers to such an extent that even the more equitable offers could not cover the cost of information. Based on our data set one cannot judge safely on causes of these unexpected results. A possible explanation is that asking for information conveys the fearful message of mistrust, attracting attention from the rational way of problem solving to getting the good result. Putting it another way, less cognitive capacity might be devoted to finding the rational way to the solution. Supporting this hypothesis, we prove that proposers with partners asking information employ more the heuristic of 'separate accounts from the two kinds of chips' and try to come at equitable allocation separately from both piles.