I do not regard Geita Sebea v. Territory of Papua [1] , as laying down any principle that would require a court, in assessing compensation in a case such as the present, to allow the unimproved value of the land plus the cost of establishing the improvements thereon less any proper deduction for obsolescence or depreciation. It seems that the observations of Starke J. [2] , and Williams J. [3] , as to the way in which land should be valued were obiter, since the matter to which they were directed does not appear to have been directly raised by the questions in the case stated (which are set out in the report [4] ). In any case, their Honours said no more than that the method there approved was suitable to be applied in the circumstances of that case, which are distinguishable from those of the present, where the improvements must be regarded as having been fragmentary and useless until formed into a finished reservoir and not, as in Geita Sebea v. Territory of Papua [1] , completely erected and ready for use. The method adopted in Geita Sebea v. Territory of Papua [1] , is familiar enough to courts concerned with valuation matters and provides a useful means of arriving at the value of a structure or improvement in many cases. However where, as in Geita Sebea v. Territory of Papua [1] , and in the present case, the relevant statute requires compensation to be assessed with reference to the value of the land acquired, evidence of the cost of improvements (whether of actual cost or of what the cost would be at the date of valuation) may be relevant to the value of the land in its improved state, but it is the value and not the cost that is the matter for ultimate determination. Some improvements increase the value of land to a greater extent than the cost, but in other circumstances the cost of an improvement may greatly exceed its value, e.g., because its wasteful design renders it unnecessarily expensive to construct, or because it is redundant or out of place and cannot be put to profitable use having regard to its situation. Moreover, the cost of a partially completed structure may exceed its value because additional expense may be occasioned by commencing construction afresh and the total cost of a structure erected in two stages may be more than if the construction had proceeded from the beginning without interruption. In the present case, where it is necessary to imagine that at the date of resumption the portion of the reservoir on the resumed land was the only part in existence, the valuation has to be made on the assumption that the respondent when it acquired the land could only get a useful reservoir by extending the walls and base on to the adjoining land. In making a valuation on the basis it would be relevant to consider evidence as to the feasibility of completing the unfinished reservoir in this way, as to any engineering or constructional difficulties that would be likely to be presented and as to what extent, if any, the costs of linking the notionally existing portion with the newly constructed portion would diminish the value of the existing structure. It was erroneous for the Land Appeal Court to regard the cost of construction as the proper measure of value; the cost was only one of the facts to be considered in arriving at the value and the weight to be given to it would depend on the other evidence, including any evidence as to the matters to which I have just referred. On the other hand, two of the learned judges in the Full Court also fell into error in saying that there was no advantage for the respondent to acquire the appellants' land (with a section of the reservoir erected on it and the remaining section not constructed) rather than to acquire vacant land and go to the expense of erecting an entirely new reservoir on that land. This observation suggests that the additional value given to the land by the section of the reservoir erected upon it may be disregarded, and such a suggestion is contrary to the principles I have already discussed.