This article presents a statistical analysis of Ethernet LAN traffic at the source level, demonstrating that high-speed network traffic exhibits self-similarity or long-range dependence, a property contrary to traditional traffic modeling assumptions. The authors propose a physical explanation for this phenomenon, based on the "Noah Effect," which refers to high variability or infinite variance in the ON and OFF periods of packet traffic. They show that the superposition of many ON/OFF sources with the Noah Effect leads to self-similar aggregate traffic, characterized by the "Joseph Effect." The study uses detailed statistical analyses of real-time traffic measurements from two Ethernet LANs, involving hundreds of active source-destination pairs, to confirm that traffic data at the source level is consistent with the Noah Effect. The results have implications for traffic modeling, synthetic traffic generation, and network performance analysis. The authors also discuss the robustness of the Noah Effect under different threshold values, showing that it remains consistent regardless of how OFF or ON periods are defined. The findings support the use of self-similar models for high-speed network traffic and highlight the importance of considering the Noah Effect in traffic modeling and network design. The study provides evidence that the Noah Effect is prevalent in both 1989 and 1994 traffic traces, indicating that the self-similar nature of network traffic is a robust feature of modern high-speed networks.This article presents a statistical analysis of Ethernet LAN traffic at the source level, demonstrating that high-speed network traffic exhibits self-similarity or long-range dependence, a property contrary to traditional traffic modeling assumptions. The authors propose a physical explanation for this phenomenon, based on the "Noah Effect," which refers to high variability or infinite variance in the ON and OFF periods of packet traffic. They show that the superposition of many ON/OFF sources with the Noah Effect leads to self-similar aggregate traffic, characterized by the "Joseph Effect." The study uses detailed statistical analyses of real-time traffic measurements from two Ethernet LANs, involving hundreds of active source-destination pairs, to confirm that traffic data at the source level is consistent with the Noah Effect. The results have implications for traffic modeling, synthetic traffic generation, and network performance analysis. The authors also discuss the robustness of the Noah Effect under different threshold values, showing that it remains consistent regardless of how OFF or ON periods are defined. The findings support the use of self-similar models for high-speed network traffic and highlight the importance of considering the Noah Effect in traffic modeling and network design. The study provides evidence that the Noah Effect is prevalent in both 1989 and 1994 traffic traces, indicating that the self-similar nature of network traffic is a robust feature of modern high-speed networks.