Distant naval bases, therefore, are vulnerable in a high degree; they are vulnerable both to direct attack and to an attack on their lines of communications; and the factors that help a base in one way injure it in another. If a naval base is placed on a rock, or a rugged little island that holds nothing else, and on which a hostile army could not land, it is very safe from land attack; whereas, if it is placed on a large and fertile island, on which an invading army could easily land, it is extremely vulnerable to land attack. But, on the other hand, the naval base on the inaccessible island could be starved out by simply breaking its lines of communications, while the naval base on the large and fertile island might be able to survive indefinitely, even though the communications were wholly ruptured. The establishment of any permanent distant naval base is a matter of great expense, even if the natural conditions are favorable. But favorable conditions have rarely existed; and the expense of establishing such bases as Malta, Gibraltar, and Heligoland has been tremendous. An important consideration has been the fact that, unless the base were made so strong that it could not be taken, it might be better not to attempt to fortify it, on the theory that it would be better to let a poor naval base fall into the hands of the enemy than a good one. To this reasoning, the answer is usually made that no base can be made absolutely impregnable, and that sufficient defense will be provided if it makes the task and cost of capturing the base greater than the base is worth. This means simply that the more valuable the base is, the more money should be spent in defending it; and that _it is worse than useless to defend it by any means that is obviously too small, in proportion to its value_.