Record and analysis require an experimental formulation of standards by which the work of government can be tested. Such standards are not to be evolved off-hand out of anyone’s consciousness. Some have already been worked out experimentally, others still need to be discovered; all need to be refined and brought into perspective by the wisdom of experience. Carried out competently, the public would gradually learn to substitute objective criteria for gossip and intuitions. One can imagine a public-health service subjected to such expert criticism. The institute of research publishes the death-rate as a whole for a period of years. It seems that for a particular season the rate is bad in certain maladies, that in others the rate of improvement is not sufficiently rapid. These facts are compared with the expenditures of the service, and with the main lines of its activity. Are the bad results due to the causes beyond the control of the service? do they indicate a lack of foresight in asking appropriations for special work? or in the absence of novel phenomena, do they point to a decline of the personnel, or in its morale? If the latter, further analysis may reveal that salaries are too low to attract men of ability, or that the head of the service by bad management has weakened the interest of his staff. When the work of government is analyzed in some such way as this, the reporter deals with a body of knowledge that has been organized for his apprehension. In other words, he is able to report the “news,” because between him and the raw material of government there has been interposed a more or less expert political intelligence. He ceases to be the ant, described by William James, whose view of a building was obtained by crawling over the cracks in the walls. These political observatories will, I think, be found useful in all branches of government, national, state, municipal, industrial, and even in foreign affairs. They should be clearly out of reach either of the wrath or of the favor of the office-holders. They must, of course, be endowed, but the endowment should be beyond the immediate control of the legislature and of the rich patron. Their independence can be partially protected by the terms of the trust; the rest must be defended by the ability of the institute to make itself so much the master of the facts as to be impregnably based on popular confidence.