AbstractA new method of measuring chromatic discrimination steps is described where observers must indicate in which direction a change has occurred, as well as the fact of its occurrence. This procedure seems to stabilize the subject's criterion of difference and yields step sizes more akin to those of Wright's “dashes” than MacAdam's ellipses. Tests have been made along two critical dimensions of variation: (1) the tritan axis, where discriminations depend initially only on variations in blue‐cone (B) excitation and (2) the red‐green axis, where discriminations depend on the substitution of red‐cone (R) excitation for that of green cones (G), or vice versa. Discriminations dependent on blue cones are affected by the level of B‐cone excitation but are independent of R/G. After individual differences are taken into account by a simple scaling factor, an equation of the form ΔB/ B + Bo = K accounts well for the data. Blue‐cone discrimination is found to be independent of the ratio of red‐to‐green‐cone excitation, with an optimal Weber fraction of about 18%. Discriminations dependent on the exchange of red‐and green‐cone excitation yield an optimal Weber fraction of about 2%. This ratio is increased, though not greatly, by an imbalance of the red‐green ratio relative to a neutral condition, and by increasing the level of blue‐cone excitation. At 120 td, the ratio between a MacAdam step and ours is approximately constant at about 13:1.

From the colorimetric considerations the following two theorems are proved mathematically.A. The maximum luminous efficiency for any colored light is achieved from the appropriate combination of two spectrum colors.B. The mixture of two spectrum colors, which gives the maximum luminous efficiency for a colored light, also gives the maximum luminous efficiency for all the colors which can be produced with combinations of these spectrum colors.A new type fluorescent lamp, named fresh white, which has a slightly greenish color has been developed on the basis of these theorems. The high output 110 W lamp of this type is as efficient as 91 1m/W. Because of its moderate color rendering property this lamp is suitable for street lighting.

けい光ランプの色温度は, その定義によってスペクトル三刺激値が変わると値が変わる. 8種類のけい光ランプ (Fig.1) の色温度を1931年 (2゜視野) と1964年 (10゜視野) とのCIEスペクトル三刺激値によって計算した (Tab.1). 一般にけい光ランプの色温度は10゜視野で低下し, また色度は黒体軌跡に対して赤紫側にずれる (Fig.2). そのずれ方は黒体放射との分光分布の相違の大きいランプほど大きい.10゜視野スペクトル三刺激値のy10λを比視感度Vλの代わりに用いると10゜視野測光量が定義できて, その最高視感度は654.5 [10゜視野ルーメン/ワット] となる. この定義によると各種のけい光ランプおよびA光源の測光量はいずれも見掛け上増加する (Tab.2).視野の大きさの変化による色温度の変化の中でも, 温白色けい光ランプの200゜K (17MRD) の変化は視感でも十分検出できる大きさであるので, これを用いてスペクトル三刺激値の実地試験を試みた. 11人の観測者 (Tab.3) が2゜および10゜の視野で温白色けい光ランプと, フィルタを掛けた霞球光 (Fig.4) との色合わせをする実験 (Fig.3) を行なった. 結果は個人差によるパラツキはあるが, 平均的にCIE標準観測者から予期される値によく一致した (Fig.5). 10゜ 視野のCIE標準観測者に対する統計データから色合わせの95%信頼長円を計算してみると, この実験における日本人観測者グループはCIE標準観測者との相違は見出せないことが結論される.なおAppendixとして10゜視野による黒体放射の輝度と色度座標の表をそえる.