distilled water, and (4) rinsing with alcohol. Before adopting this cleaning procedure, representative specimens of low, medium, and high gloss enamels were scoured with calcium carbonate for 650 strokes. No change in gloss was noted on the high gloss enamels while an increase of about 5 percent was noted after 350 strokes on both the medium and low gloss enamels. The gloss increased by about 15 percent after 650 strokes. The increase caused by the scouring may be the result of a smoother surface which reflects light better than the original surface. Since the increase was only 5 percent after 350 strokes (enough for 10 inspections) it was thought that this treatment would not significantly affect the exposure test results. The scouring treatment produced no noticeable color change.

the boiling acid solubility and citric acid spot test ratings for the enamels included in this test are presented in table 1.

2.6. Thickness

The enamel thickness was measured at four locations on each specimen with an eddy current thickness gage [4]. The average thickness for each enamel is presented in table 1.

2.7. Gloss and Color

The 45-degree specular gloss of the enamels was measured [5] at four orientations near the center of the specimen before and after exposure. The average initial gloss for each enamel is given in table 1. The gloss after exposure is reported as the percentage gloss retained.

The color is measured before and after exposure with a color difference meter [6]. One of the three average storage specimens of each enamel was selected as the color standard to obtain the maximum efficiency possible with this type of instrument. The color change was calculated [7] from the color values and is reported as color retention which is 100 minus the color change in NBS units.

3.1. Visual Observation of Enamels

a. Amount of Soil Retained

A visual examination revealed that enamels exposed at New York City had the most dirt adhering to them while those exposed at Los Angeles, Washington, Montreal, and Kure Beach had respectively cleaner surfaces. The dirt deposits were readily removed from the enamels exposed at all sites except New York by the cleaning procedure outlined in section 2.4. The enamels exposed at New York required vigorous scouring with a sponge that was sprinkled with calcium carbonate to remove the dirt film.

b. Spalling

After the specimens were cleaned it was observed that "fishscale" type spalling occurred on some enamels exposed at all sites. In no case was the spalled area large enough to constitute failure [8].

c. Visual Evaluation of Color and Gloss The enamels exposed at Washington, New York, Montreal, and Los Angeles appeared to have good color and gloss retention with the exception of the red enamels, which appeared lighter than the unexposed enamels. Many of the enamels exposed at Kure Beach had obviously undergone substantial color changes. The reds, blacks, dark greens, and blues lightened considerably. The red, dark green, and blue enamel surfaces had the appearance of light particles

surrounded by a darker colored matrix, while the black enamel surfaces appeared uniformly lightened. Many of the white enamels exposed at Kure Beach appeared to be yellowing. The weathering of some of the enamels exposed at Kure Beach had progressed to a point where changes in the gloss were readily apparent and the surfaces appeared chalky.

3.2. Instrumental Determinations of Color and Gloss

The average color- and percentage gloss-retention values obtained from the six-months', one-year's and three-years' inspections of these enamels are presented in tables 2 and 3. The data in figures 1 and 2 indicate that gloss does not change as uniformly with exposure time as color. The nonuniformity is prob ably caused by the exceptionally large percentage changes which result from small gloss changes for the low gloss enamels. For example, a change of one gloss unit will produce a 17 percent change in the gloss retained for low gloss enamel, AR-C, but only a 1 percent change for high gloss enamel, AO-A. The increase in percentage gloss retained for many of the low gloss enamels after three years' exposure (see table 3) may be caused by a smoothing of the surface by the weather.

An increase in the gloss-rentention values of many low and medium gloss enamels exposed at New York City was also observed. These increases are attributed more to the severe scouring treatment necessary to

clean these specimens than to the action of the weather. Since the percentage gloss-retention values were more variable than the color-retention values, the remainder of this report will be based primarily on color retention.

3.3. Comparison of Exposure Sites

The color-retention data were subjected to a twosided sign test [9] to determine if the exposure conditions at the various sites caused significant differences in the color retention of the enamels. The exposure sites were divided into three categories on the basis of the observed results of the sign test. Kure Beach was in a category by itself and caused the greatest color change in the enamels. New York and Washington were grouped together and caused less severe color changes than Kure Beach but more severe than the sites in the third category, Montreal and Los Angeles.

3.4. Effect of Different Colors

The average color retentions for the nine enamel colors are presented in table 4 and figure 3. It can be seen that large color changes occurred for the red enamels exposed at all sites. This was expected since all the red enamels included in this test failed the nitric acid test for color retention [10] and would not be recommended for outdoor exposures.

Large color changes also occurred on the black, dark green, grey, and blue enamels exposed at Kure Beach. These changes were not predicted by any of the current tests for porcelain enamels on aluminum.

3.5. Effect of Initial Gloss

The enamels included in this test program were produced in three gloss ranges; low, having an initial 45-degree specular gloss of 35 or less; medium, between 36 and 69; and high, 70 or over. The average color-retention data (top section of table 5) indicate